[proof-of-concept] Visualization of curved meshes with VTK + ParaView

[sloede]

This is just to demonstrate how to visualize a structured curved mesh with VTK + ParaView. The intent is to have a simple, working prototype for @bennibolm @erik-f @NichtLucas that can be extended to supersampling (= better resolution) and 3D. If you have questions, please let me know.

Usage

Run the waving flag elixir:

julia> using Revise, Trixi

julia> trixi_include("examples/2d/elixir_advection_waving_flag.jl")

Then open the resulting file curved.vts in ParaView. This should result in something like the following (grid lines shown for illustrative purposes):

[sloede]

@bennibolm @erik-f @NichtLucas I just added some comments to the implementation for your benefit. Right now, it is a very hacky way of visualizing the structured DG solution in VTK: The original DG nodes are interpreted as the vertices of quadrilateral, straight-sided VTK cells, and the mean of the four vertices is used as the value of the enclosing cell.

This approach extends straightforward to 3D and should allow for an initial, minimum effort visualization of 3D results.

In the intermediate term, it would make sense to

• first interpolate to a uniform visualization mesh, i.e., similarly sized visualization nodes in each element
• allow for supersampling, i.e., using more visualization nodes (= VTK cells) than DG nodes in each element (right now, we even use even less...)

I think these two extensions are also relatively easy to implement if one is familiar with all the interpolation routines etc. If you're up for it, we can sit together and talk through it.

[sloede]

Not relevant anymore.