Complex networks and agent-based models

Centre of Excellence in Computational Complex Systems Research


The science of complex networks is inherently interdisciplinary, which is also reflected in our research interests. On one hand, we do fundamental research on complex networks, which involves theoretical work, modelling and method development, usually inspired by empirical data on networks. On the other hand, we apply the "machinery" of network theory to various problems from a wide range of sciences, typically in collaboration with research groups at LCE and elsewhere.

Here are some topics we are currently working on:

Weighted complex networks.

Typically, networks represent interacting systems, such that the vertices depict the interacting elements and the edges their interactions. It is evident that in many cases, for a better understanding of the systems in question, the nature and strength of these interactions should be taken into account. This is readily done by assigning weights to the edges of the networks. These weights can represent e.g. the strength of social ties or flows in networks of chemical reactions. Evidently, when the "simple" network picture becomes more complicated when a new degree of freedom is added, and new tools and methods have to be developed for characterizing and measuring the properties of networks. This is also a practical problem; much of our method development has been related to research where the weights and their relation to network structure carry a lot of information of the system under study - this information just has to be extracted.

For more, see, e.g.

Structure and function of social networks.

Social networks are fascinating - we are all part of an enormous network of social ties, spanning the entire planet, but our visibility is more or less limited to at most a few hundreds of people. It is intriguing to study the structure of social ties on very large scales, up to millions of people - how is the microscopic structure reflected in the macroscopic properties of social systems? Here, modern-day electronic databases contain a wealth of information. Recently, we have been studying a large dataset related to mobile telephone communications between a large number of individuals. By applying weighted complex networks methodology, we have been able to confirm some sociological theories, as well as discover a number of new, exciting facts. In addition to empirical data analysis, social networks and "sociodynamic" processes (spread of information, opinion formation under peer pressure, etc) can also be studied with simulations and theoretical models, drawing inspiration from empirical observations.

For more, see, e.g.,

Financial and economic networks.

In today's world, everything is becoming more and more dependent on everything else, especially in the global economy, as evident from stock market crashes quickly reverberating around the globe. The global economy and its various "subsystems", such as stock markets, are also approachable from the complex network point of view. Networks can represent "direct" financial interactions, such as the international trade between the world's countries, as well as more abstract, indirect interactions such as those between stock prices, inferred from financial time series. One of our targets is to better understand how to extract meaningful information from these correlations, using the weighted complex networks framework as well as spectral analysis.

For more, see, e.g.,

Networks in biology and evolution.

The networks group at LCE participates in the EU-funded research project EDEN, which was launched in January 2007. In this project, we approach the genetic relationships of marine plants with network methods on several levels, from the Tree of Life to networks of gene flows between and within local populations. For more, see the EDEN website.