Is it possible to get neighbour cell or points? about pyvista-support

[pswpswpsw]

@pswpswpsw, could we add your blog post to the Long Examples section of the website?

Also, is there a status on the proposed feature requests? Should we outline some features that should/could be added to vtki?

No problem! Indeed, I will have some more fancy figures later within 2 months on that post! I will keep you updated!

[akaszynski]

I’m not familiar with what a large stencil is, but I’d probably go with using a kdtree as you’ll be able to quickly query the tree once you’ve set it up. pykdtree is one of the best out there, and it carries no dependancies.

If you do need me to setup an adjency function, please comment with your intended pseudo-code.

[akaszynski]

Still need help on this?

[supersubscript]

I agree that it would be nice to have a few different adjacency/neighbourhood functions. These examples come to mind for neighbouring elements on a mesh:

• https://www.vtk.org/Wiki/VTK/Examples/Cxx/PolyData/VertexConnectivity
• https://www.vtk.org/Wiki/VTK/Examples/Cxx/PolyData/CellPointNeighbors
• https://www.paraview.org/Wiki/VTK/Examples/PolyData/CellEdgeNeighbors

And then this for a point cloud:

• https://www.vtk.org/Wiki/VTK/Examples/Cxx/DataStructures/KdTreePointLocator/ClosestNPoints

Would certainly be features that'd be useful to have.

[pswpswpsw]

Still need help on this?

it is not urgent for me but I think other people might be interested.
Btw I wrote a blog on postprocessing OpenFOAM data using VTK and then use your package for various purpose in Python:

So the pseudo-code I have in my mind would be

---

let's say we tried to find 2-level of Neighour for each cell

• for loop over all the cells, say i-th cell for instance:

  • find the neighbour of i-th cell (I think VTK itself might have already done it for you), denote as set J_i

    • for loop over each cell in set J_i, for each cell, find the neighour of that cell, denote as J_{J_i}.

  • for each cell Let's do a union operation of all the set J_{J_i}.

Then This union set for i-cell is the 2-level neighour set. Note that we know all of the spatial information so we can do really fancy things like computing any partial derivatives on any directions, so getting curvature is trivially easy.

[akaszynski]

This is different than a meshless kdtree as it requires knowledge of the mesh connectivity. I’d be happy to add in cell based neighbor querying.

BTW, I liked the blog. If you wish, you can also visualize within python without heading over to paraview with:

grid.Plot(scalars=p2_cell)

See help(grid.Plot) for more details on how you can change the color map and other plotting variables.

[banesullivan]

@pswpswpsw, could we add your blog post to the Long Examples section of the website?

Also, is there a status on the proposed feature requests? Should we outline some features that should/could be added to vtki?