Duplicated points (nodes) in exported VTU file, no connectivity used?

[lamBOOO]

I noticed that the number of nodes printed by gmsh (when reading a .msh-file for example) doesn't match the "number of points" displayed in Paraview's information tab. The same behavior happens if we use the built-in CartesianDiscreteModel.

The below code shows, that exporting a (2,2)-quad triangulated unit square leads to a VTU file, which has some nodes duplicated. Instead of the required 9 nodes (see attached image), we get 16 nodes.

I think that instead of using connectivity information, the writevtk(Ω, "trian")-function simply places every element inside the "world". In contrast to this, the writevtk(model, "model") works as expected.

Is this behavior intended? Couldn't we save a lot of memory by using the connectivity information as well and not writing nodes multiple times?

Or do I have a fundamental error in thinking about unstructured meshes? 😄

using Gridap
domain = (0, 1, 0, 1)
partition = (2, 2)
model = CartesianDiscreteModel(domain, partition)
reffe = ReferenceFE(lagrangian, Float64, 1)
V = TestFESpace(model, reffe)
Ω = Triangulation(model)
f = interpolate(x->x[1] + x[2], V)
writevtk(Ω, "trian")
writevtk(model, "model")

using ReadVTK
readtrian = VTKFile("trian.vtu")
readmodel = VTKFile("model_2.vtu")

get_points(readmodel)
# 3×9 reshape(reinterpret(Float64, view(::Vector{UInt8}, 9:224)), 3, 9) with eltype Float64:
#  0.0  0.5  1.0  0.0  0.5  1.0  0.0  0.5  1.0
#  0.0  0.0  0.0  0.5  0.5  0.5  1.0  1.0  1.0
#  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0

get_points(readtrian)
# 3×16 reshape(reinterpret(Float64, view(::Vector{UInt8}, 9:392)), 3, 16) with eltype Float64:
#  0.0  0.5  0.0  0.5  0.5  1.0  0.5  1.0  0.0  0.5  0.0  0.5  0.5  1.0  0.5  1.0
#  0.0  0.0  0.5  0.5  0.0  0.0  0.5  0.5  0.5  0.5  1.0  1.0  0.5  0.5  1.0  1.0
#  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0

# FIXME: The triangulation VTU has duplicated points, is this intended?

I use Julia 1.8.2 and Gridap 0.17.16.

[oriolcg]

I think this is an intended behavior as it enables, for instance, displaying discontinuous fields.

[lamBOOO]

Thank you. Yes, good point, I was not thinking about discontinuous fields. 🤔

[fverdugo]

Yes, the visualization grid is indeed independent from the interpolation grid. This provides extra flexibility (e.e. discontinuous fields).