A new tutorial for Field.mesh()

[LSchueler]

We should add a tutorial showing how to use the Field.mesh method to generate random fields on pre-existing meshes.

[MuellerSeb]

You are totally right! I'll take a look at this next week.

[banesullivan]

It'd be awesome to have this tutorial use PyVista 😉

[LSchueler]

Why not provide 2 or more tutorials? - Maybe you already have something like a minimal working example?
I can also beef it up a little with text, if you just want to provide the mesh file and the code.

[MuellerSeb]

There are now some tutorials with MeshIO: #112. PyVista examples still missing, but the interface is now ready: #59

[MuellerSeb]

ogs5py has its own tutorial: https://geostat-framework.readthedocs.io/projects/ogs5py/en/stable/tutorial_02_pump.html

[MatusGasparik]

Even though this issue has been already closed, what I am still missing is a tutorial with the Field.mesh() + pyvista in conjunction with the Field.transform() functionality. I am not quite sure what is the best way to use the two APIs together...

Calling the Field.mesh() with an instance of the pv.Grid as an arg, saves the field into the pv.Grid. What if you then want to generate some transformations of the original field using Field.transform() and also want to have these saved in the pv.Grid as well?

Inspired by the example for discrete fields, this is how I got it to work:

import gstools as gs
import pyvista as pv

dim, spacing = (30, 20, 10), (.1, .1, 0.1)
grid = pv.UniformGrid(dim, spacing)

model = gs.Gaussian(dim=3, len_scale=[2, 1, 0.1])
srf = gs.SRF(model, seed=19970221)

srf.mesh(grid, points="points", name="random-field")          # stores "random-field" scalar array in the `grid` object

vals = np.linspace(np.min(srf.field), np.max(srf.field), 5)
srf.transform("discrete", field="field", store="discrete", values=vals)     # stores "discrete" field in `srf` object

grid['discrete'] = srf['discrete']          # copies data from `srf` to `grid`


plotter = pv.Plotter(window_size=(800, 600))
plotter.background_color = 'gray'
plotter.enable_anti_aliasing()
plotter.add_mesh(grid, scalars="discrete", show_edges=True)
plotter.show()

Result:

Is this how those two APIs are supposed to be used together? It would make sense to call srf.mesh after all the transformed fields have been defined in, or can subsequent calls of srf.mesh be used to add new fields to the pv.Grid object?

[banesullivan]

Just a drive-by thought/comment:

After calling, srf.mesh, srf could keep a pointer to the mesh and add fields to it anytime an operation is performed. Otherwise, grid['discrete'] = srf['discrete'] is exactly what you'd have to do.

---

I also want to give a plug for an example I put together here: https://banesullivan.com/pyvista/examples/kriging.html

[MatusGasparik]

I also want to give a plug for an example I put together here: https://banesullivan.com/pyvista/examples/kriging.html

Kudos for that link to your blog website! Looks fantastic. Added to my reading list.

[MuellerSeb]

As @banesullivan pointed out, you can add more fields afterwards. Closing this for now. If you have further ideas on how to improve the interoperability, just open another issue or discussion.