Skip to main content
. 2016 Jul 22;6:30108. doi: 10.1038/srep30108

Figure 2. Laplacian-based Network Embedding.

Figure 2

(a) A network of 750 nodes was generated by means of the PS model, with target average node degree 2m = 10, scaling exponent γ = 2.75 and network temperature T = 0. The network is embedded to the hyperbolic plane Inline graphic with LaBNE to reveal the angular position of the nodes in the hyperbolic circle containing the network. (b) Finally, the radial coordinates of the nodes are assigned, so that they resemble the rank of each node according to its degree. By the colour of the nodes, which highlights their angular coordinates, one can note that the embedding by LaBNE is rotated by some degrees with respect to the actual node angular coordinates obtained with the PS model. This does not impact the hyperbolic, distance-dependent connection probabilities, because distances are invariant under rotations. Edges in the raw embedding by LaBNE are not shown for clarity.