Matt
Thanks. Indeed, setting "Parent" to the list of parents is working fine.
Cactus vertex-centred data do not look nice, as the refinement level interfaces are clearly visible.
I attach two plots, one for cell-centred AMR, the other for vertex-centred AMR. Both show r, the radial coordinate, and were produced by
p = yt.SlicePlot(ds, "z", "GRID::r")
p.set_cmap(field="all", cmap="plasma")
p.set_log("all", False)
p.set_zlim("all", 0, 2)
p.save("GRID-r.png")
Both datasets have three levels 0,1,2, where level 1 covers (almost) all of level 0, and level 2 ranges has approximate size [-20,20]^3.
-erik