Berlin 2015 – scientific program
DY 59.4: Poster
Thursday, March 19, 2015, 16:00–18:00, Poster A
Significance tests for topological characteristics of spatially embedded networks — •Marc Wiedermann1,2, Jonathan F. Donges1,3, Reik V. Donner1, and Jürgen Kurths1,2 — 1Potsdam Insitute for Climate Impact Research, Germany — 2Humboldt University, Berlin, Germany — 3Stockholm Resilience Centre, Stockholm University, Sweden
Spatially embedded complex networks, i.e., networks where the nodes are embedded in some metric space, have attracted increasing attention in many fields of science. In such systems, it is of particular interest to study which network properties can solely be explained by the spatial embedding of the nodes and which are unique to the system under study with respect to a predefined null hypothesis. For this purpose, we introduce a set of random null models for spatially embedded networks, which are constrained by the spatial embedding of the nodes to different extents. These surrogate networks are generated by randomly shuffling edges in the original network but preserving spatial properties such as the edge length distribution or the average length of edges emerging from each node. For three different real world systems, we use our framework to evaluate to what extent measurable network characteristics can be explained by the spatial embedding of the system alone. The proposed random network models serve to generally evaluate the significance of links and network characteristics in general spatially embedded networks and are applicable also to functional networks, where links represent significant similarities between different localized areas or monitoring points in the system under study.