search

MESAHub / mesa

Modules for Experiments in Stellar Astrophysics
http://mesastar.org
GNU Lesser General Public License v2.1
145 stars 39 forks source link

TDC fails on ns_he. #312

Open adamjermyn opened 3 years ago

adamjermyn commented 3 years ago

The ns_he test case fails with TDC on. The latest commit where I’ve checked this is c351d55.

I’ve been working on this the last few days and my best theory is that the initial model (generated with make_env) is too far out of thermal equilibrium for TDC to handle. Or else the accretion luminosity turns on too suddenly for TDC to handle. In either case looking at the mixing window, MLT is able to rapidly spawn some strong convection zones to get into thermal equilibrium in the first tens of steps whereas TDC struggles because it has to gradually turn them on. The latter is of course more physical…

I confirmed that running for 100 steps on MLT then restarting with TDC gets around this issue on c351d55. I haven’t checked if it runs all the way to the end of the test case, but I ran through the start of helium burning up to a luminosity of logLHe=4.9, which isn’t far from the termination condition.

I’ve also done some basic sanity checks on TDC with this test case to confirm that it is behaving as intended on gradT/Y/conv_vel/D, including that it agrees with Cox MLT on the long-time-scale solution.

So I think this is either a problem with the accretion turning on suddenly or else with the initial model.

I don’t have the timeto track this down further so I’m just turning TDC off on this test case, but if someone with an interest in neutron star accretion wants to look further please do!

adamjermyn commented 3 years ago

I verified this on 6b537f2: TDC causes the case to hit lots of retries and fail.