I wonder if it is implemented in networkx.drawing to hold one dimension during the layout optimization with a predefined position array.
Lets say you want to optimize the layout of a graph and have the x dimension of the positions already given, so you only want to optimize the y directions of the vertices.
So far I've only noticed, that one can hold positions of certain vertices, but then of course non of those are being moved.
In the Python package grandalf, they have DicgoLayout, so I'd expect something similar in networkx.
I ended up with forking the spring layout with a hold_dim parameter, so during updating the positions only x or y is being changed.