Guillaume Broggi

In the last case the shapes are attached to the same body, so the I was (maybe naively :) ) expecting that the shape copies would reference the same body...

From a user perspective, it would make sense to have a "ready to use" copied object, with all referenced objects. But I see that it would be very challenging from...

In this case it seems that batched adding/removing bodies might not be worth the effort :) > Anyway, I think it could be possible to make an array based version...

Hi Farzad, This is likely to be a Unix related issue, like bash is not installed or not where it should be. I can confirm that the instructions are working...

Hi @xwpken, could you elaborate on the correct format of `internal_vars_surfaces`? For instance, if there is one `get_surface_map` expecting one scalar to update the load?

Thanks @xwpken ! I managed to obtain the proper shape with: ```python def get_surface_maps(self): def traction(u, x, load): return np.array([0.0, 0.0, load[0]]) return [traction] ``` and ```python shape = (len(problem.boundary_inds_list[0]),...

However, this behavior makes sense, for instance, for concentric tubes. ```python mesh = pv.Tube(radius=2) + pv.Tube(radius=3) + pv.Tube(radius=4) mesh.plot(jupyter_backend="notebook") binary = mesh.voxelize_binary_mask(dimensions=(20, 20, 20)).points_to_cells() binary.plot(jupyter_backend="notebook") ```  

I have included these examples in the docs: https://github.com/pyvista/pyvista/pull/6680 If the behaviour for internal surfaces is expected, then one could take the boolean union of the two surfaces to include...

Examples added to the doc by #6680

Thanks for the investigation. Could it be improved with #5610?