Complex Systems Summer School 2012-Project Presentations

Complex Systems Summer School 2012 Home Surveys/Evaluations Logistics Faculty Participants Schedule Lecture Readings Background Readings Software Projects & Working Groups Tutorials Blog Glue Session Forum Project Presentations Final Papers After Hours Photos TShirts

Use this space to post project presentations and outlines. Include group members, a brief outline, and your slides.

Price-time Dynamics of Contracts Traded on Prediction Markets

Joanne, Vikram, Matteo, Sanith

Prediction markets have been shown to outperform traditional methods of polls and opinion surveys in forecasting future events. The futures contracts traded in these markets assess the expectation of occurrence of a variety of events spread across multiple domains (political, economic, entertainment, financial and weather). We explore the feasibility of 'predicting' the outcome of binary true/false prediction market contracts ahead of their expiry date using a neural-network based machine learning approach. In addition we focus on the characteristics of political-based contracts to establish whether they exhibit characteristic 'fundamental' properties.

How Complex Languages Replicate through Simple Brains

Katrien, Vanessa, Sandro, Cameron, Jasmeen

Through the use of an iterated learning experiment, we investigated the transmission of a "high entropy", randomised initial language through successive generations of participants. We want to see what features of the language replicated most easily, and what structure emerged by the end of the chain. Our hypothesis is that the language converges to a "low entropy" equilibrium state with a minimal number of words, morphemes, and form-meaning distinctions.

Collaboration in times of stress: an Agent Based Modelling approach

Fabio Cresto Aleina, Elena del Val, Tom Fennewald and Friederike Greb

We want to investigate the influence of exogenous stress on cooperative behaviour. We propose an agent based model in which the agents can be interpreted as farmers living in a water limited environment. With changes in precipitation patterns, the farmers undergo stress, and we observe how this impacts relationships among farmers and their properties.

Simple variation of the logistic map as a model to invoke questions on cellular protein trafficking

(Sepehr Ehsani, http://arxiv.org/abs/1206.5557)

Many open problems in biology, as in the physical sciences, display nonlinear and 'chaotic' dynamics, which, to the extent possible, cannot be reasonably understood. Moreover, mathematical models which aim to predict/estimate unknown aspects of a biological system cannot provide more information about the set of biologically meaningful (e.g., 'hidden') states of the system than could be understood by the designer of the model ab initio. Here, the case is made for the utilization of such models to shift from a 'predictive' to a 'questioning' nature, and a simple natural-logarithm variation of the logistic polynomial map is presented that can invoke questions about protein trafficking in eukaryotic cells.

Changes in Social Network Structure in Response to Crisis: Using Twitter data to Explore the Effect of the Tōhoku Earthquake.

Christa Brelsford and Xin Lu

Abstract: We use twitter data from 7 days before and after the Tōhoku Earthquake to explore how cooperation rates, social network structure and connectivity, and social network vulnerability changed in Japan in response to the disaster. An English language data set is collected for the same time period to use as a control. Data is collected for a period of 96 hours starting from March 4th 2011 2:46pm JST and for 96 hours beginning March 11th 2011 2:46 pm JST. The rate of cooperative behavior, measured by the occurrence of helping words in tweets increases slightly in the English dataset and by an order of magnitude in the Japanese dataset. A network analysis is also performed. Network edges are retweets and direct messages. In future work, we hope to explore whether problem solving capacity in a social system changes in response to crises, based on changes in the rate of cooperation and information transfer in a network.

The CSSS Network

Tom & Riccardo (with JP and others)

We will investigate the questions you are dying to know: What interesting interactions are revealed from the first 3 weeks of the Complex System Summer School survey? Have barriers between academic disciplines been broken down? Do power laws fit the data!? ...

Let us know if you have specific questions or if you would like to be involved in data analysis!

Is there a method in the madness? the dynamic structures of human language use

Priya and Riccardo

Psychiatric anecdotal reports point to the monotony, lack of emotion and sometimes intelligibility in many clinical populations. Linear measures of fluency and prosody, however, present only controversial differences between patients and healthy controls and only in unnatural phonations (i.e. say "aaaaa" for 30 secs). We therefore go complex and chaotic on a set of more ecological recordings and transcriptions from 4 clinical populations (Asperger's, Schizophrenics, Depressed and Right Hemisphere Damage patients) as well as from healthy controls. We then set a classifier-driven race: will non-linear analyses outcompete linear analyses in discriminating between pathologies?

Escaping the Poverty Trap: Modeling the Interplay Between Economic Growth and the Ecology of Infectious Disease

Georg, Ben, Laurent, Oscar

The dynamics of economies and infectious disease are inexorably linked: economic well-being influences health (sanitation, nutrition, etc) and health influences economic well-being (labor productivity lost to sickness and disease). Often societies are locked into "poverty traps" of poor health and poor economy. Here, we demonstrate poverty traps formed in models of infection and endogenous growth, as well as ways to break out of poverty traps. We explore two mechanisms of escape: one, through an influx of capital, and another through changing the percentage of GDP spent on healthcare. We find large influxes of capital is successful, but increasing health spending does not. Our results have important policy implications in the distribution of aid and within-country healthcare spending.

The Targeting and Timing of Treatment Influences the Emergence of Influenza Resistance in Structured Populations

Ben, Laurent, Oscar, Georg

Evolution of antiviral resistance in influenza carries large societal impacts through morbidity and mortality caused by treatment failure. Several previous studies put forth theory regarding the optimal timing, targeting and absolute level of treatment in populations. Few of these studies, however, have considered populations with explicit structure. Here we present a model of antiviral resistance on networks and explore the timing, targeting and levels of treatment. Interestingly, we find bistability as a result of treatment leading to the existence of an unstable manifold, above which successful treatment (i.e.: no resistance) is impossible. We find, contrary to previous results, that degree-targeted treatment is not optimal, and leads to higher levels of resistance than random treatment. Additionally, in accordance with previous results, we find an optimum level of treatment which is less than 100%. These findings findings have important consequences in guiding policy behind influenza treatment. The bistability indicates that caution should be taken when treating populations when the absolute numbers of infections are unknown. Positively, our results indicate that putting resources into targeted treatment is not necessary, random treatment is preferable.

Learning in Random Boolean Networks

Nick A., Keegan, Matteo, Vikram, Sarah, Mark

Inspired by biochemical networks which adapt on evolutionary timescales, neural networks that adapt during development and learning, and universal computation in cellular automata, we have implemented several models of learning in Random Boolean Networks (RBNs) in order to better understand the relationships between network structure, node interaction rules, and network output.

Enzyme Catalysis and the Outcome of Chemical Reactions

Piotr and Georg W. Enzymes are catalysts that accelerate chemical reactions but do not affect their outcome. This traditional paradism was developed under artificail test tube conditions. Our project investigates the possibility that the presence of an enzyme can alter the course of a reaction if it takes place under more physiologic conditions.

How Does a Stochastic Environment affect Community Assembly?

Xue, Χλοε, Xiaoli

We are interested in how exogenous temporal variability in resource availability affects the community structure of organisms with different resource-use strategies. Organisms induce additional resource stress on each other through competition. This is an abstraction of an arid environment with unreliable rainfall; the organisms themselves have been abstracted to four unitless parameters that allocate their resources to different parts of their lifecycles. The system has memory, as the previous presence of an organism affects the resource transport mechanism (an abstraction of soil).

How Does a Network’s Structure Influence its Traceability?

Xin and Abby

We are interested in systematically studying the problem of finding the source of a contamination spread through a network. We model a contamination spreading through the food distribution network, which we represent by interconnections between farmers, distributors, and retailers, and construct an estimator for the outbreak source given only this structure. We show how the ability of the estimator to narrow down the source identification problem changes with the connectivity and the number of observations. We propose a measure for traceability, or the overall ability to identify the source of spreading given any set of outbreak observations, based on entropy. We show how this measure appropriately reflects the range of uncertainty in identifying the source. We believe this measure may be useful in the design of networks that are conducive to accurate identification of the source of contamination.

We Got the Skills to Pay the Bills: Exploring the Link Between Occupation Diversity and Innovation

Andrés, Charlie, Gareth, and Nick

Where does diversity in skills or occupations come from and why does it lead to more innovative cities? Previous work in this area has shown that there is a scaling behaviour which allows citizens of larger cities to earn an extra 15% in income whilst making use of 15% fewer gas stations, for example. Making use of occupation, patent, and population data of US Metropolitan Statistical Areas (MSA), we try to understand what factors make successful cities. Here we assume that successful cities are those cities which are most innovative as determined by the production of patents. In addition we use agent-based modelling to explore how and why people acquire new skills and whether this leads to more productive cities.

Space of complex networks and robustness

Ian, Marco, Xin, and Oleksandr

Complex networks have various properties which can be measured in real networks (WWW, social networks, biological networks), e.g. degree distribution, modularity, hierarchy, assortativity etc. Robustness of complex networks is a big question, however only some progress have been done in this direction. For example, it was shown that the scale-free networks are much more topologically robust to random attacks than random networks. Various characteristics of complex networks might influence the robustness differently. The question is how?

We generated continuous topological space of networks with respect to degree distribution (from random to scale-free) and clustering (from none to high). Then we attacked the network by removing vertices randomly and highly connected (hubs). The next step is to calculate network robustness, it is non-trivial task because there are many different ways to do it. So far we calculate the size of giant component during attack process for the entire space.