search

nschloe / pygmsh

:spider_web: Gmsh for Python
GNU General Public License v3.0
867 stars 162 forks source link

Dimension header when saving a 2D mesh #97

Closed r-maas closed 7 years ago

r-maas commented 7 years ago

Hi Nico,

Thanks for creating such useful tools as pygmsh and meshio. I have been using them for a few days now, and it has been a big help. I did run into a small problem when I tried to create a 2D mesh. I understand that gmsh requires 3D points, and the "dim" variable in the generate_mesh function here is passed correctly to gmsh. However, if I save the output geometry as a .mesh file, it always seems to list "Dimension 3" regardless of "dim".

If I first manually adapt the array containing the points and then save the mesh meshio does correctly store it as "Dimension 2":

out = pg.generate_mesh(geom,dim=2)

points=out[0][:,0:2]
cells=out[1]

meshio.write(
    'test_forced_2D.mesh',
    points,
    cells,
    )

However, this doesn't seem very elegant. Am I doing something wrong, or is there maybe a better way of ensuring that a 2D mesh is saved with the correct value for the Dimension?

Best regards, Ruben

nschloe commented 7 years ago

Neither pygmsh nor meshio have explicit support for 2D meshes. Perhaps that'll be something to add? I've never come across a case where a zeroed-out third dimension would be a problem. For this, I'll have to check which mesh formats actually support 2D meshed. I assume there won't be many.

r-maas commented 7 years ago

Ok well if it's not a feature it's not a bug! So I'll close this issue.

Sfepy gave an error when I tried to load a .mesh file with Dimension 3 for a 2D finite element problem. I'll check how the mesh is read in.

nschloe commented 7 years ago

Feel free to reopen if this continues to be a problem for you.

gdmcbain commented 7 years ago

The specification of the MEDIT format by Frey (2001, §7.2.1, p. 33) only allows Dimension 3. That does seem slightly odd: if so, why bother with the line? But that's my reading.

Another program (besides SfePy reported by @r-maas above) FreeFem++ wants its MEDIT .mesh files for two-dimensional problems to have:

Since Gmsh always thinks in three dimensions and writes MEDIT files with Dimension 3, I wrote a small 3to2.awk script which I include in my Make/Shake pipelines between (Py)Gmsh and FreeFem++. It's available at https://bitbucket.org/gdmcbain/freefem and has also been posted numerous times to the freefempp and Gmsh mailing lists; e.g.: