search

svalinn / Cubit-plugin

Plugins and command extensions for Coreform Cubit
BSD 3-Clause "New" or "Revised" License
17 stars 14 forks source link

Debugging MCNP import into Cubit #155

Open rmchurch opened 2 months ago

rmchurch commented 2 months ago

I can use the import of MCNP .i file into Cubit for some files, but there is one that is causing a segfault. I have looked for ELL, WED, ARB flags, but it doesn't have them. Are there additional ways to debug? I started to build from scratch mcnp2cad, but figured I would first see if there are additional steps recommended to find root cause.

MicahGale commented 2 months ago

This isn't a way to debug the problem inside of the Cubit plugin, but may still be helpful. I am wondering: are you sure your input file is valid? A quick plug for another utility: MontePy (I am the lead developer) can do quite a few sanity checks that your input file is valid for MCNP 6.2+, and in my experience sometimes more aggressive than MCNP.

pip install montepy
montepy -c <foo.i>
rmchurch commented 2 months ago

Thanks, thats useful. I was told this ran successfully on MCNP 6.2. I tried out montepy but got an error for a line: 

530   35  -2.334 -60066.2 -60077.3 -60088.1 60077.1 -61 201 vol=3.53E+07 imp:n,p=1

where it doesn't like the floating point:

ValueError: invalid literal for int() with base 10: '-60066.2'
makeclean commented 2 months ago

That looks like it doesn't like the macrobody facets

On Wed, 2024-06-05 at 10:51 -0700, Michael Churchill wrote: valid literal for int() with base 10: '-60066.2'

makeclean commented 2 months ago

Can you share your input? One thing I would try is remove half the cell definitions inthe input deck, if it still fails remove a further half, and keep bisecting until you find the problem surface.

rmchurch commented 2 months ago

Attached. I'll start trying the bisection method.

Base2.i.txt

MicahGale commented 2 months ago

@makeclean you are correct that this is an issue that MontePy can't handle macrobodies. Can this plugin handle Macrobody facets?

makeclean commented 2 months ago

If memory serves, then yes I'm fairly sure it does

makeclean commented 2 months ago

Looks like its the 2nd cell which is the first problem... - intersection of three tori

makeclean commented 2 months ago

I think our implementation doesn't actually have the correct logic for this case; we missed the case where the major radius -s less than 0, we have the case for major radius < minor radius, but not the third case - I'll raise a bug