Project Details
Projekt Print View

Dissecting evolving interaction networks: Which network components are important for the dynamics?

Subject Area Statistical Physics, Nonlinear Dynamics, Complex Systems, Soft and Fluid Matter, Biological Physics
Term from 2017 to 2022
Project identifier Deutsche Forschungsgemeinschaft (DFG) - Project number 391934314
 
Final Report Year 2022

Final Report Abstract

Complex networks can be found in animate and inanimate nature on virtually all scales. Examples include interconnected cellular structures (e.g., the brain), communication (e.g., the Internet) and infrastructure networks (e.g., air traffic, power grid), social, and even climate networks. Interdisciplinary research into complex networks has advanced from a characterization of network properties on various scales to improving our understanding of collective dynamical phenomena by means of time-evolving interaction networks. Still, disentangling the complicated relationship between structure and dynamics remains a challenge, mostly because progress in identifying key constituents (nodes and links) of time-evolving interaction networks is limited by a number of factors. With our research programme, we targeted at improving (1) the identification of constituents that are important for the network’s structure and dynamics and (2) the characterization of properties of such key constituents. By means of application-oriented method developments, we modified various, widely-used and well-known centrality concepts for nodes to those for links, in order to find which links (or groups thereof) in a network are important between other pairs of nodes. With these concepts, importance of network constituents for structure and dynamics of the larger network can be assessed from different perspectives. We could verify the suitability of our novel approaches with computer simulations of a number of paradigmatic model networks under controlled conditions and could demonstrate their usefulness through applications to various real-world networks. With an eye to the analysis of empirical networks, where knowledge about and access to key network constituents is usually very limited, we investigated how well this information can be derived from incomplete observational data of a network’s dynamics. With quite extensive computer simulations, we identified possibilities and limitations of a data-driven identification of key constituents as well as characterization of other, more global network properties. Eventually, we employed our newly developed analysis concepts and methods to identify and characterize key constituents in so called functional brain networks from subjects with epilepsy. We derived these networks from long-lasting (days to weeks) electroencephalographic recordings that cover various dynamical regimes of the human brain and achieved novel insights into network mechanisms underlying epilepsy and the generation of seizures. These insights significantly contribute to further improve prediction and control of seizures. With our research findings, we could contribute to advancing our understanding of local aspects of the relationship between structure and dynamics in natural and man-made evolving interaction networks.

Publications

 
 

Additional Information

Textvergrößerung und Kontrastanpassung