|
First Summer School on Aspects of Complexity July 18-28th, 2005 |
|
|---|
|
First Summer School on Aspects of Complexity July 18-28th, 2005 |
|
|---|
Summary:
We examine problems that arise in certain technological network structures that have gained recent notoriety and practical importance. These structures, which include Peer-to-Peer (P2P) and mobile ad hoc networks (MANETs), are characterized by their extreme dynamism and large scale. Problems that need to be solved include those related to implementing functions such as routing, searching, power management, load balancing and distributed computation. We survey a class of techniques for solving these problems that are inspired by biological systems or proceses. The resulting functions are not only typically very simple, but are often completely decentralized, extremely robust, self organizing and scalable. In addition to gaining an insight into the formalism behind the techniques, the students will have an opportunity to gain practical insight through hands-on experimentation using the PeerSim simulation framework.
Summary:
I will describe a simulation-based model of goods and of the production and exchange of goods. The model is focused on the two great transitions in the manner in which goods are produced in human societies: (1) the transition from societies in which families are the only existing type of organization for the production of goods to societies in which families exist together with a centralized structure, the state, that produces and redistributes new types of goods, and (2) the transition to societies in which families are replaced by a plurality of a new type of organizations for the production of goods, private enterprises, which produce and sell the goods they produce.
Summary:
The main objective of this lecture is to introduce an alternative approach in modeling financial markets and economics in general. Standard economic theory emphasizes the role of rational agents who maximize utility. Sometimes the economic agents' aggregate strategic behavior is dominated by the constraints of the market institutions. In this case the rationality assumptions can be relaxed or even dropped. Outline: (i) Introduction and empirical evidence in financial markets, (ii) A zero intelligence model of financial markets based on the limit order book; (iii) Model validation using high frequency data from the London Stock Exchange, results and model limits; (iv) Zero intelligence model as starting point/benchmark, new assumptions on agents' behavior, evaluation of the effect of the new assumptions on the statistical properties of prices and market efficiency; (v) Heterogeneity is added to the model: agents are differentiated on information and strategic specialization, (vi) Artificial market tournament.
The World Wide Web (WWW) is the result of spontaneous and uncoordinated decisions made by people around the world, and yet it displays some striking regularities. Perhaps even more surprisingly these regularities can be observed in networks arising in very different contexts, such as protein interaction networks, the internet (not to be confused with the WWW), and even many kinds of social networks. We will explore some of these regularities looking at some of the experimental work done. Then we will briefly discuss some mathematical models that have been introduced to explain the structure of these networks. Finally, we will focus on some computer science aspects of the WWW showing how these regularities can be exploited to improve the performance of search engines. In particular, we will explain the basic mechanisms underlying some very popular search engines such as Google. While this last topic requires a certain mathematical competence, the rest of the discussion should be accessible to a very large audience.
Summary:
Most optimization problems have been known for centuries, think of the Chinese postman problem, first formulated by Euler in 1736. These problems have the characteristic of being combinatorial, that is, all the possible combinations of the decisions and variables must be explored to find a solution. The downside of this situation is that as the number of decisions and variables increase the time required to find a solution becomes rapidly unaffordable. Heuristics methods have been devised to rapidly explore only parts of the search space, thus reducing the time required to obtain a solution, which is often sub-optimal, but already a good improvement from the starting situation. A heuristic makes use of peculiar characteristics of a problem and exploits them to find a solution. Other empirical methods do not exploit only the problem characteristics but especially the analogy with other optimization methods found in Nature. One of the most recent and powerful heuristic is Ant-Colony Optimization (ACO). ACO is based on the observation that real ants find the optimal path between a food source and their nest food by depositing chemical traces (pheromones) on the floor. A computer analogy has been implemented where a large number of simple artificial ants are able to build good solutions to hard combinatorial optimization problems via low-level based communications based on artificial pheromone. We present the basic ACO principles already published in prestigious journals like Nature, Scientific American and Harvard Business Review and we show the most recent applications used to optimize hard logistic industrial problems.
Summary:
Description In the first part of the lecture an overview of different sources and types of networks derived from textual data will be presented, such as: citation networks, collaboration networks, words co appearance networks, ontologies and dictionary networks, news analysis networks (KEDS - Kansas Event Data System). In the second part some approaches to analysis of such networks will be described and illustrated on selected networks. For example, recently developed islands algorithm can be used to identify 'themes' in a given network. Applied to US Patents network (~ 4 millions of vertices, ~ 16 millions of arcs) it revealed the 'liquid crystal display' as the main theme, but also many others. All these approaches are supported by program Pajek. See also the data sets.
Summary:
The goal of this section is to introduce students to the theory and practice of computer based simulations with agent based models; in this perspective a key part oh the session is the afternoon laboratory with Guido Fioretti. Issues that we will be addressing include: (i) Why and how to use simulation and agent based models in social science; computer simulation as a language for (social) science; simulation and model specification, versatility, efficiency; (ii) The NetLogo, Swarm and JAS platforms; (iii) Agent based model of artificial stock market; agent based models of enterprises and organizations. Use and development of a multi model tool for micro-macro interaction, and (iv) From the prey-predator model to structures of interaction and coevolution in workers-firms models and banks-firms models.
Summary:
This course offers an introduction to computational modeling with applications
to the social sciences. Thanks to advances in the natural sciences and the
decreased cost of computer technology, computational modeling is becoming
increasingly popular as a tool in the social sciences. This course focuses
on agent-based modeling, which is a particular type of computational methodology
that allows the researcher to create, analyze, and experiment with artificial
worlds populated by agents that interact in non-trivial ways. In these "complex
adaptive systems", computation is used to simulate agents’ behavior
and cognitive processes in order to explore emergent macro phenomena, i.e.
structural patterns that are not reducible to, or even understandable in
terms of, properties of the micro-level agents. Such "bottom up" models typically
feature local and dispersed interaction rather than centralized control.
Moreover, as opposed to traditional models that either assume either a small
number of dissimilar or numerous identical actors, agent-based models normally
include large numbers of heterogeneous agents. Rather than studying equilibrium
behavior, the focus is often on dynamics and transient trajectories far from
equilibrium. Finally, instead of assuming the environment to be fixed, many
agent-based models let the agents constitute their own endogenous environment.
Existing applications in the social sciences include neighborhood segregation,
social stratification, artificial stock markets, ethnic conflict, party formation,
balance of power politics, and cooperation among democratic states. The
goal of the course is to introduce the participants to the principles of
agent-based modeling. No previous knowledge of programming is required but
is of course helpful. Programming will not be taught, but a tutorial introducing
the tools will be given in the last session. See Repast, the leading open-source
simulation package for agent-based modeling in Java.
Summary:
Developments in multi-agent based simulation have offered a new way of doing sociology: by conducting virtual experiments on artificial societies. In this lecture, I shall review the recent achievements of such computational sociology, comment on the implications for sociological methodology, and consider whether this way of doing sociology can make any impression on its 'big questions', such as understanding social stratification, culture and power.
Living systems are more complex in function and structure than anything we know from the inanimate world. This complexity arises from the large number of competing, reinforcing, and substituting regulatory mechanisms that operate over a broad range of different time and space scales and from the large number of hierarchical levels from the level of the individual ligand or receptor molecule over organelles, cells, functional units and organs to the full body and its response to external signals. Living systems also operate under far-from-equilibrium conditions, and they display complex nonlinear dynamic phenomena in the form of oscillations, bifurcations, chaos, synchronization, etc. The challenge to physics and mathematics is to develop the concepts and methods we need to deal with this level of complexity and to collaborate with the biologists in order to apply the methods and concepts in a proper manner.
The talk will illustrate some of the types of complex dynamics we meet in living systems and show how mechanism-based modeling can be used to understand some of these phenomena.
The services available to consumers over the web often run on heterogeneous platforms, retain heterogeneous definitions, and have heterogeneous implementations. The objective of this lecture is to develop and experiment a foundational methodology capable of describing service specifications and supporting a discipline for their aggregation. This methodology relies on services as the fundamental elements for developing applications, thus conforming with a novel paradigm called Service Oriented Computing (SOC).
| Application: | Before June 1st, 2005 |
|---|---|
| Payment & Registration: | Within one week upon notification of acceptance |
| Refunding in case of cancellation: | Before July 1st, 2005 (net of processing fee) |
| Summer School: | July 18-28th, 2005 |
Bertinoro itself is picturesque, with many narrow streets and walkways winding around the central peak. The meeting will be held in a redoubtable episcopal fortress that has been converted by the University of Bologna into a modern conference center with computing facilities and Internet access. From the fortress you can enjoy a beautiful vista that stretches from the Tuscan Apennines to the Adriatic coast.
| Scientific Committee | Erik R. Larsen, City University of London |
|---|---|
| Alessandro Lomi, University of Bologna | |
| Local Organization | Guido Fioretti, University of Bologna |
| Andrea Bandini, Ce.U.B. | |
| Elena della Godenza, Ce.U.B. | |
