Percolation of interlayer feature-correlated multiplex networks

Abstract

Multiplex networks formed by a set of nodes connected through different types of edges are widely used to model interactions between different systems. In many real-world scenarios, the properties of nodes in different layers are characterized by non-topological features such as age, flow, and geographical location. On this basis, the associations between different layers are caused by the coupling of features. Here, we propose a percolation framework to investigate the properties of such interlayer feature-correlated multiplex networks (FCN). Based on this framework, the robustness of the networks correlated through completely unequal discrete features, discrete features with repetition, and continuous features is analyzed. Theoretical and numerical results show that the interlayer degree–degree correlation (IDDC) of the network is controlled by both the correlation between degrees and features, and the correlation between the features of different layers. When there are repeating features, the network structure is randomized due to the decrease in the differentiation of nodes in the feature space. This randomization substantially reduces the IDDC of the network even if the above correlations are fixed. Our findings provide new insight into revealing the origins of the complexities of real-world multiplex networks.