Figure 1 | Scientific Reports

Figure 1

From: Looking beyond community structure leads to the discovery of dynamical communities in weighted networks

Figure 1

Schematic representation of the method to determine and validate an approximate cluster partition for a network. The clustering algorithm of “Section The case of nearly equitable clusters” is fed with data on a network (in particular, its adjacency matrix) and a given number of clusters, k. This algorithm determines an approximate cluster partition for which we calculate the corresponding projection operator \(E_H\). In turn, this is used to calculate \(\delta A\) from Eq. (6) with \(R=E_H\). The correction cost \(\psi =\Vert \delta A\Vert\) is then contrasted for statistical significance with the result of randomized graphs obtained from the original network, swapping the links.

Back to article page