Luca Cardelli (web site)
Biological Systems as Reactive Systems
(Tuesday, 9:50 -10:40)
Systems Biology is a new discipline aiming to understand the behavior of biological systems as it results from the (non-trivial, "emergent") interaction of biological components. We discuss some biological networks that are characterized by simple components, but by complex interactions. The components are separately described in stochastic pi-calculus, which is a "programming language" that should scale up to description of large systems. The components are then wired together, and their interactions are studied by stochastic simulation. Subtle and unexpected behavior emerges even from simple circuits, and yet stable behavior emerges too, giving some hints about what may be critical and what may be irrelevant in the organization of biological networks.
Henrik Lund (web site)
Modular Robotic Systems as Complex Systems
This talk presents the robotic building block concept in which processing and physical aspects of robotic artefacts is distributed. The technological concept of physical building blocks with processing, input, output (incl. communication) is derived from embodied artificial intelligence that emphasises the role of interplay between morphology and control. The building block concept is exemplified with a variety of applications in toys, self-assembling robots, and playware. Playware is the use of intelligent technology to create the kind of leisure activities we normally label play, i.e. intelligent hard- and software that aims at producing play and playful experiences among users. We developed the modular tangible tiles as components for a new kind of playground, on which children can experience immediate feedback on their motions. Hence, this kind of playground allows implementation of games and plays that demand physical activity amongst the users, and thereby contribute as a new tool in the fight against obesity. The tangible tiles are homogenous robotic building blocks, which gives assembly, substitution and production advantages. Also, studies show that using neural networks it may be possible to classify children’s behaviour, and use such classification to develop adaptive playgrounds.
Bio : Henrik Hautop Lund, full professor, the Maersk Mc-Kinney Moller Institute for Production Technology, University of Southern Denmark. His research group, the AdapTronics group (www.adaptronics.dk), focuses on research in modern artificial intelligence. He is member of the Danish National Research Council (FTP), has published more than 90 peer-reviewed papers, and has initiated and coordinated numerous, big research projects such as HYDRA on self-assembling robots (www.hydra-robot.com), Body Games, Intelligent Artefacts, Flexible robots for SMEs, VIKI Humanoid (RoboCup Humanoids Freestyle World Champions 2002), Playing with Ambient Intelligence, etc. He is co-inventor and co-founder of RoboCup Junior. He has collaborated with numerous companies such as LEGO, Bandai, Siemens, ABB, Microsoft. His robot work has been presented to e.g. prime ministers, HM Queen of Denmark, and HM Emperor of Japan.
Giorgio Parisi (web site)
The complexity from a point of view of a physicist
In this talk I will try to express my view point on why complexity is interesting for physics and why the physical approach is interesting to complexity, A few example will be presented, stressing both the accomplished results and the the open problems, The importance of interdisciplinary research in this context will be stressed.
Peter Schuster (web site)
The complexity of genotype-phenotype maps and its consequences for evolution
(Wednesday, 9:00- 9:50)
Genotype-phenotype mappings are indispensable for a comprehensive understanding of evolutionary optimization. At the current state of the art only one example of a genotype-phenotype map is sufficiently well known in order to use it as a basis for modeling evolution: The sequence-structure map of RNA molecules, which can be investigated also experimentally, for example through the evolutionary design of RNA aptamers using the SELEX technique. Several evolutionarily relevant features of this mapping from sequence space into a space of structures, representing the phenotypes in RNA evolution experiments, have been discovered in the past: (i) High degree of neutrality leading to connected neutral networks spanning whole sequence space, (ii) shape space covering predicting that each common RNA structure can be reached from (almost) everywhere in sequence space through a fairly small number of point mutations, and (iii) the intersection theorem, which states that there exists at least one sequence for any arbitrarily chosen pair of structures that can fold into both.
Modeling evolutionary optimization by computer simulation of replication, mutation, and selection of RNA molecules in a flow reactor revealed several characteristic features, among them: (i) The optimization process occurs in steps rather than continuously because fast adaptive phases are interrupted by long quasi-stationary epochs of neutral evolution, (ii) the evolution of a population in the stationary epochs corresponds to a diffusion process on a neutral network, and (iii) the exploration of sequence space by replication and mutation can be interpreted as a kind of primitive learning at the population level. This learning process has many features in common with the foraging strategies of ant colonies.
The simple scenario assigning one structure to a given sequence becomes more realistic and more complex through considering suboptimal structures and explicit folding kinetics of RNA molecules. Comparing the set of suboptimal structures with the set of structures in the one-error neighborhood of a neutral network, called the shadow, allows for modeling the evolution of Boltzmann ensembles at different temperatures. Interaction between RNA molecules through cofolding provides another dimension of higher complexity within the RNA model. Several examples of multi-conformational RNA molecules will be presented. Such molecules have been designed and occur also in nature where they have important regulatory properties in molecular genetics.
Peter Schuster, Institut für Theoretische Chemie der Universität Wien, Austria
Peter Schuster, Molecular insight into the evolution of phenotypes. 2003. In: James P. Crutchfield and Peter Schuster, eds. Evolutionary Dynamics - Exploring the Interplay of Accident, Selection, Neutrality, and Function. Oxford University Press, New York, pages 163-215.
Walter Fontana and Peter Schuster. 1998. Continuity in evolution. On the nature of transitions. Science 280:1451-1455.
Peter Schuster and Walter Fontana and Peter F. Stadler and Ivo L. Hofacker. 1994. From sequences to shapes and back: A case study in RNA secondary structures. Proc.R.Soc.Lond. B 255:279-284.
Wolf Singer (web site)
The brain seen as a goal-oriented, self-organizing complex system
Our intuition assumes that there is a centre in our brain in which all relevant information converges. This, so our introspection, would be the place where the signals provided by our various sensory systems are bound together into coherent percepts of the surrounding world, where decisions are reached, where plans for future acts are elaborated and where adapted motor responses are programmed. Eventually, this distinguished site would be the place where the intentional self constitutes itself.
The results of neurobiological investigations design a radically different picture. The brain presents itself as a highly distributed system in which a very large number of processes occur simultaneously and in parallel without requiring coordination by a central convergence centre. The connectivity graph of the brain is characterized by a dense network of mainly reciprocal links between the chains of processing nodes that span between the sensory and executive organs. This specific architecture resembles in many respects small world networks and raises the question, how the multiple operations occurring in parallel are bound together in order to give rise to coherent perception and action.
Based on data obtained from investigations of the visual system mechanisms will be discussed that could accomplish the binding of distributed processes into coherent representations. The hypothesis will be forwarded that temporal coherence serves as an important organizing principle and that this coherence is achieved by the synchronization of oscillatory activity in distinct frequency bands.
Tamas Vicsek (web site)
Uncovering the overlapping community structure of complex networks
in nature and society.
(Tuesday 9:00- 9:50)
Many complex systems in nature and society can be described in terms of networks capturing the intricate web of connections among the units they are made of. A fundamental question of great current interest is how to interpret the global organisation of such networks as the coexistence of their structural sub-units (communities) associated with more highly interconnected parts. Identifying these unknown building blocks (e.g., functionally related proteins, industrial sectors, groups of people) is crucial to the understanding of the structural and functional properties of networks. The existing deterministic methods used for large data sets find separated communities, while most of the actual networks are made of highly overlapping cohesive groups of nodes. Here we introduce an approach to analyse the main statistical features of the interwoven sets of overlapping communities making a much needed step towards the uncovering of the modular structure of complex systems.
After defining a set of new characteristic quantities for the statistics of communities, we apply an efficient technique to explore overlapping communities on a large scale. We find that overlaps are indeed very significant, and the distributions we introduce reveal novel universal features of networks. Our studies of collaboration, word association, and protein interaction graphs demonstrate that the web of communities has highly non-trivial correlations and specific scaling properties.
*This works has been carried out in collaboration with G. Palla, I. Derenyi
and I. Farkas
Douglas White (web site)
Multi-net analysis and nonlinear dynamics: some methods and results in complexity science
(Wednesday, 9:50 -10:40)
Many complex systems are composed by multi-nets, i.e., multiple networks undergoing change in time series. Understanding the behavior of multi-net systems poses some basic questions:
1) how should we represent and model multiply layered and evolving networks (multi-nets) so as to discover their instabilities and nonlinear dynamics?
2) what are some of the common properties induced by dependence on co-evolution with network topologies?
3) does a generalized Boltzmann-Gibbs entropy, that takes into account network dependencies and hence long-range correlations, have applicability to modeling complexity in social systems?
4) what is the contribution of a combination of multiply layered networks, time series, methods of study for nonlinear dynamic interactions (identifying oscillations and instabilities), simulation, nonextensive BG entropy, and tracking co-evolution of network topology?
The examples illustrated are city attributes and networks, industrial networks, agent search behavior, and marriage choice; each includes issues of the co-evolution of network topology and micro-macro linkages. Five sets of results are discussed:
1. A simulation that shows how modeling of results with generalized Boltzmann-Gibbs (q-) entropy takes long range correlations into account in known network dynamics relating to agent search behavior. http://arxiv.org/abs/cond-mat/0508028
2. A q-entropy worldwide scaling of the 28 historically estimated city-size distributions is investigated for nonlinear instabilities in urban systems.
3. Investigation of a multi-net coding and longitudinal analysis of agrarian civilizations as dynamical networks (Medieval European and Eurasian datasets) showing nonlinear dynamic interactions.
4. Analysis of collaborative multi-nets in the world biotech industry shows an interactive dynamics of recruitment for innovation and organizational consolidation. AJS 210(4): 1132-1205.
5. Multi-net construction of social structure through mate choice and co-evolution of social network topologies. Complexity 8(1):72-81.
Jean-Christophe Yoccoz (web site)
A panorama of the mathematical theory of dynamical systems.
( Monday 9:50 -10:40)
A dynamical system consists of a phase space together with an evolution law. The goal of the theory is to understand the long term behaviour of the system. The study of stationary regimes is organized around two poles : quasiperiodic behaviour, with a rotation on a circle as paradigmatic model, and chaotic-hyperbolic behaviour, with the doubling map on the circle as paradigmatic model.
|Supported by the Future Emerging Technologies programme of the information Society Technologies programme of the European Commission.|