gmsh2nek is creating #v002 .re2 file

[Hashnayne-Ahmed]

I think this is a bug that needs to be adjusted in the gmsh2nek.f90 file that produces #v002 .re2 file instead of #v003.

[stgeke]

@yhaomin2007 Can you please have a look.

[AlexandreTlili]

Hi,
I encountered the same issue as @Hashnayne-Ahmed. As a consequence, I am not able to run a Nek simulation with a .re2 file generated with gmsh2nek in the current latest git version (commit 2c81113 from May 15 2023).
However, I can still run Nek with re2 files generated with gmsh2nek from the V19 website release.
This seems to have something to do this the version of the .re2 file (#v002 or #v003).
Do you have any advice ?
Best,

[yhaomin2007]

hi, can you be more specific about what error you see? thanks

[stgeke]

@yhaomin2007 I think we discussed this before. The correct version to use is v003. Not sure why change it some time ago.

[AlexandreTlili]

hi, can you be more specific about what error you see? thanks

@yhaomin2007
When using the latest git version of Nek to both create the .re2 file with gmsh2nek (v002) and run my code, I have Vanishing Jacobian errors with 3D simulations. Same when I use the V19 release to both create the re2 file (v003) and run my code. In both cases, more than half of my grid cells generate the Jacobian error, despite using a valid structured mesh.

The solution which seems to work in my case, is to create the v003 re2 file with the V19 release, then use the latest git commit (2c81113) to compile and run my code. (I hope it can help other people facing the same issue.)

A colleague of mine tried to run my code with a previous git version, and was able to use my re2 file, no matter if it was v002 or v003. Maybe it is due to a problem of re2 versions compatibility when reading the file at the beginning of the simulation, and going back to v002 in gmsh2nek could solve this.