Regensburg 2007 – scientific programme
Parts | Days | Selection | Search | Downloads | Help
AKSOE: Arbeitskreis Physik sozio-ökonomischer Systeme
AKSOE 16: Traffic Dynamics, Urban, and Regional Systems
AKSOE 16.5: Talk
Friday, March 30, 2007, 11:30–12:00, H8
Global traffic analysis reveals geographical modules across political boundaries — •D. Brockmann1, F. Theis1, and L. Hufnagel2 — 1MPI for Dynamics and Self-Organization, 37018 Göttingen — 2Kavli Institute for Theretical Physics, Santa Barbara, CA 93106
Geographical boundaries are key determinants of various spatially extended dynamical phenomena. Examples are migration dynamics of species, the spread of infectious diseases, bioinvasive processes, and the spatial evolution of language. As political boundaries have become less important, it is difficult to quantify their impact on spatially extended human dynamics. The evolved complexity of contemporary human travel may exhibit intrinsic modularities and effective boundary structures, which not necessarily coincide with existing political boundaries.
We investigate a large scale complex network of human travel between the approx. 3000 counties in the US. We construct the network by analyzing the flux of over 10 million dollar bills reported at the bill-tracking website wheresgeorge.com which extends the dataset of a previous study (Brockmann et al., Nature 2006) by a factor of 20. We investigate to what extend geographical information is intrinsically encoded in the topology of the network by applying two graph cutting algorithms (Newman&Girvan, Phys.Rev.E 2004 and an extension of Lee&Seung, Nature 1999) in order to identify effective clusters. Although both algorithms employ two completetely different paradigms they identify approx. 10, nearly identical effective clusters in the network. Surprisingly, these clusters are spatially compact regions, although both algorithms have no prior knowledge of geographical information. Most importantly, the geographic boundaries between the clusters only partially overlap with the political state boundaries. We conclude that graph cutting algorithms can efficiently determine effective clusters in geographically embedded transport networks. The results may aid the development of models for dynamical phenomena evolving on these networks.