Subsidieronde Complexiteit open voor vooraanmeldingen

14 oktober 2009

Het NWO-thema Dynamica van complexe systemen, kortweg Complexiteit, initieert maatschappijgedreven multidisciplinaironderzoek op het gebied van complexe systemen. Het onderzoek richt zich op systemen van verschillend karakter, zoals complexe processen bij infrastructuurnetwerken, de fluctuaties in aandelenkoersen en veranderingen in klimaat. Het thema richt zich op een brede onderzoeksgroep, waaronder ook gedragswetenschappen en economie. De subsidieronde is open voor het indienen van vooraanmeldingen. De deadline is 1 december 2009. 
Ga naar de subsidiepagina van Complexiteit.

..............................

Voorbeelden van Complexiteit-onderzoek

Brain research
There is now increasing evidence that anatomical and functional connectivity networks in the brain display the small-world phenomenon and have degree distributions with ‘heavy tails’ (Stam and Reijneveld, 2007). However, most studies show that brain networks deviate in significant ways from the classical scale-free model of Barabasi and Albert. In particular, fMRI studies have suggested a degree distribution with power law and exponential components. Such a degree distribution can be explained by growth models of brain networks, which take into account biological constraints. Furthermore, there is increasing evidence that the complex structure of brain networks breaks down in various neurological and psychiatric disorders such as Alzheimer’s disease, brain tumours, epilepsy and schizophrenia. Modern network theory provides a general framework for understanding how different types of network damage (‘random error’ versus ‘targeted attack’) could bring about these pathological network changes. Relevant research questions in the context of brain research are: What are the emergent properties of different kind of neural network structures? Can we understand the properties and limitations of macro processes, such as working memory, from the complex emergent properties of neural activity? How do emergent properties of neural activation such as awareness, influence this neural activity? But other questions also arise. One cause (say, child abuse) has often many different effects, and one effect (say depression) can have many different origins. Complex system theory may provide insight in this type of relationships and points to new types of interventions that are not based on simple cause-effect models. Similar considerations hold for understanding neurological and psychiatric disease from a complex systems point of view.

Traffic management
Transport is another discipline witnessing a shift to multi-scale modelling. The interest of transport modellers involves different time horizons. First, traffic-flow prediction is focused on traffic flows during the day or even segments of the day. Second, transport demand models are focused on longer term prediction of transport demand as a function of demographic and economic change, jointly with the impact of infrastructure, institutional, land use and economic policies. Traditionally, traffic-flow forecasting has been based on operations-research allocation algorithms applied to aggregates of travellers. Lately, however, scientists have started applying agent-based micro-level simulation models, with emerging, nonlinear aggregate traffic-flow patterns (see e.g., Balmer et al, 2005; Rosetti et al, 2005). An interesting aspect is that travellers can actively decide what to do if traffic-flow predictions are available. Different situations may emerge – multiple user equilibrium, bifurcation or oscillatory behaviour. For transport demand-modelling activity-based models are the state-of-the-art. They simulate which activities are conducted where, when, for how long, and with whom. These models depend on data collected for typical days. Only very recently, dynamic models have been developed, these models are based on notions of multi-scale interactions, nonlinear dynamics, agent-based technology, and emerging aggregate behaviour. They incorporate both processes of gradual change, but also sudden bifurcation and phase transitions.

Financial markets
The stock market presents yet another example of a highly unpredictable system. According to the traditional view financial investors are fully rational and stock markets perfectly efficient. In such a world movements in stock prices are only driven by random news about the economy (e.g. interest rates, economic growth, etc.). But the standard view is at odds with extreme movements observed frequently in financial markets worldwide, for example the 20% drop of the Dow Jones index on black Monday, October 19, 1987 or, more recently, the large movements due to the credit crisis. In the last decade an alternative view based upon complexity theory has been proposed. The most prominent example has been the Santa Fe artificial stock market (Arthur et al., 1997), where the interaction of a large population of fundamental traders and technical analysts leads to large swings in stock prices triggered by news but reinforced by herdingbehaviour. Using a simpler, stylized version of this model, Brock and Hommes have shown that evolutionary selection among simple investment strategies may lead to instability and chaotic stock price fluctuations, with temporary bubbles and unpredictable crashes, very similar to real markets (see Hommes 2006). In fact, agent-based financial market models can reproduce important ‘stylized facts’ or emergent properties observed in real financial data, such as unpredictable asset returns, clustered volatility (i.e. irregular switching between quiet and turbulent phases) and fat tails.