| Contents |
| We use an edge-switching algorithm to produce random realisations of a particular network degree distribution. These networks, we claim are maximum entropy realisations. That is, they are random realisations of a given class of networks. When we apply this to the particular case of scale-free complex networks and we find a richer variety of complex networks than typified by (for example) preferential attachment. For weighted (and unweighted) complex networks we define a path-length depedent measure of variability in the structure of the network: this quantity we call the network entropy - as it is defined in an entropy-like manner based on the homogeneity of path-lengths between random nodes. We study the variation in this property across network classes and apply it to the special case of networks generated from time series via an ordinal partitioning of the quantised scalar signal. |
|