C# Класс SharpNeat.View.Graph.DepthLayoutManager

Layout nodes based on their depth within the network. Note 1. Input nodes are defined as being at layer zero and we position them in their own layer at the top of the layout area. Any other type of node (hidden or output) node not connected to is also defined as being at layer zero, if such nodes exist then we place them into their own layout layer to visually separate them from the input nodes. Note 2. Output nodes can be at a range of depths, but for clarity we position them all in their own layer at the bottom of the layout area. A hidden node can have a depth greater than or equal to one or more of the output nodes, to handle this case neatly we ensure that the output nodes are always in a layer by themselves by creating an additional layer in the layout if necessary. Note 3. Hidden nodes are positioned into layers between the inputs and outputs based on their depth. Note 4. Empty layers are not possible in the underlying network because for there to be a layer N node it must have a connection from a layer N-1 node. However, in cyclic networks the output nodes can be source nodes, but we also always paint output nodes in their own layout layer at the bottom of the layout area. Therefore if the only node at a given depth is an output node then the layout layer can be empty. To handle this neatly we check for empty layout layers before invoking the layout logic.