[Paraview] partitioned vtu and polyhedral cells

Marco Cisternino marco.cisternino at optimad.it
Wed Sep 16 10:45:36 EDT 2015


Good morning,
I write the result of my parallel simulation on octree grids in vtu files (one per process), only using hexahedra, and I collect them in a pvtu file.
My data are collocated on the cell centers.
In order to use the Contours filter of Paraview, I firstly apply the CleanToGrid filter, then the CellDataToPointData filter and at the end the Contours one.
By this way I can avoid discontinuous contours at process boundaries, but I can't avoid contour discontinuities at grid level jumps.

In order to overcome this problem, I tried to write my grid using polyhedra/hexahedra at grid level jumps and only hexahedra elsewhere.
If I open any single vtu file, I can see all the cells I wrote in that file. But on a single vtu mixing polyhedra and hexahedra if I apply the following pipelines
CellDataToPointData -> Contours
or
CellDataToPointData -> Slice
the last filter show the result in the hexahedra.
What happens inside polyhedra?

Moreover, If I open the pvtu file I cannot see polyhedra, even if asking paraview for grid information I get the right number of cells.
What about polyhedra in this case???

Lastly, I decided to do something working in any case. I open my pvtu with only hexahedra.
Therefore I call the following pipeline,
CleanToGrid -> CellDataToPointData -> Delaunay 3D -> Contours
The result is not discontinuous, but at grid level jumps my contours oscillate. These ones are better than discontinuous ones, but ....

Any suggestion about how to get continuous and smooth contours in cell-centered octree grids are appreciated.

Thanks for your attention,

Marco Cisternino
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/paraview/attachments/20150916/4627b8bc/attachment.html>


More information about the ParaView mailing list