[jtrin]

Hello!

I'm attempting a shape design with a few geometric constraints using shape_optimization.py. My simulation runs fine until direct CFD is run on my_mesh_deform.su2 in DSN_002. Then I get a segfault right after ParMETIS partitions the graph.

Here's a quick screenshot of the segfault message.

I'm running on 60 cores with (nominally) 384GB of memory available. So I feel I should have more than enough memory available. I also tried running parallel_computation.py using the deformed mesh, and I get the same error.

I noticed that the deformed grid seems very different from the initial grid. It seems like a lot of volume cells have gone missing. Screenshots before deform, and after.

Is it possible that deformation merged some of the grid points? I took a look at the surface and volume deformations and didn't see much out of the ordinary but it might be hard to tell from inspection.

Any ideas as to how I might resolve this? I'm hoping it's just a simple error in my config file. Thanks.

[talbring]

Hi Jason,

I just opened a Pull Request that fixes that problem. It was introduced in v7.0.1, so v7.0.0 should work fine ... Next maintenance release (end of this month) will include that fix.

[jtrin]

Originally Posted by talbring

Hi Jason,

I just opened a Pull Request that fixes that problem. It was introduced in v7.0.1, so v7.0.0 should work fine ... Next maintenance release (end of this month) will include that fix.

Great, thank you!

[jtrin]

Update: installing v7.0.0 does indeed resolve this issue. Thanks again!