sisl.viz Grid plot uses Grid.geometry instead of Grid.lattice

[zerothi]

Describe the bug

The problem occurs when the lattice is different from the Geometry.lattice.

Reproducable code

lattice = H.geometry.lattice.untile(3, 0)
grid = sisl.Grid(0.2, geometry=H.geometry, lattice=lattice)
grid.plot()

[pfebrer]

I didn't know it was even legal to pass both a geometry and a lattice to Grid :sweat_smile:

[pfebrer]

I don't know if I should fix this or keep it in mind for the sisl.viz PR. My plan is to work on that during August so that I can finally finish it and merge it to make sisl.viz developeable again.

[zerothi]

We can wait, but perhaps it would be nice if we could raise an error if users do this? I.e. if the lattice is different from the geometry.lattice, where could I add this?

[pfebrer]

What is the error exactly? Is it that when you use plot_geom=True you get the lattice of the geometry?

[zerothi]

If I do grid.plot() it shows the lengths as though it was the geometry.lattice but it should show the grid.lattice, i.e. the values are correct, but it gets stretched and axis are wrong. I wanted to raise an error in the code if they are in-commensurate.

[pfebrer]

Ah ok, you can raise the error here:

https://github.com/zerothi/sisl/blob/1ef16ed437174d84c510f6872f01f0613ffcbe9a/src/sisl/viz/plots/grid.py#L539-L545

Using self.grid or grid. But does different lattices make sense for other sisl functionalities?

[zerothi]

Ah ok, you can raise the error here:

https://github.com/zerothi/sisl/blob/1ef16ed437174d84c510f6872f01f0613ffcbe9a/src/sisl/viz/plots/grid.py#L539-L545

Using self.grid or grid. But does different lattices make sense for other sisl functionalities?

Yes, when you do grid.sub() or expand densities to grids, then a different lattice is nice (consider graphene -> square grid)

[pfebrer]

I see, yes it makes sense :+1: