GRUPO INTERDISCIPLINAR DE SISTEMAS COMPLEJOS
DEPARTAMENTO DE
MATEMÁTICAS
UNIVERSIDAD CARLOS III DE MADRID
Avda. Universidad, 30 – 28911 Leganés, Madrid
Tfno. 916 249 411 Fax 916 249 129 [email protected]
http://gisc.uc3m.es/~anxo
Scientific Report
Working Group 4 “Evolution and Co-evolution”
(Meeting code: COST-MP0801-260109-05429)
Kick-Off Meeting
Madrid, January 26-28, 2009
The meeting took place at Hotel Carlton, Madrid, organized by Anxo Sánchez (Universidad
Carlos III de Madrid, Spain, coordinator of WG4). All COST-supported participants stayed at
the Hotel thus maximizing the time available for the meeting discussions and networking
activities. The Hotel provided a meeting room with all the necessary facilities plus the catering
(meals and coffee breaks). Everything took place smoothly and all the participants were
satisfied with the organization.
The meeting was the kick-off mini-workshop of Working Group 4 “Evolution and Coevolution” of COST Action MP0801 “Physics of Competition and Conflict”. It was intended as
a discussion forum where researchers from different fields exchange views and put their ideas in
common to identify important/promising directions for research in this area. The focus was on
the relevance of the evolutionary paradigm in socio-economic phenomena. Different
perspectives were considered by participants from several backgrounds and fields: Physics,
Mathematics, Economics, Sociology and Biology.
Topics discussed in depth at the meeting included (the discussion leader is indicated in
parenthesis): models of coevolution of different populations/opinions under mutual interactions
(Jensen); modelling of social networks obtained from field data such as questionnaires about
friendship relationships (Steglich); models of coevolution of economic strategies and networks
from theoretical and computational viewpoints (Fosco); the interplay of majority rules for
evolution of opinions or social traits with the thresholds for their propagation (Galam); the
interpretation and modelling of data on languages and their origin and relationships in terms of
evolution as compared with evolution of biological species (Manrubia); models of cooperative
behavior under weak and strong selection (Traulsen); coevolution of networks and strategies in
social dilemmas (Moreno); migratory phenomena and their relation to the emergence of
cooperative behavior in societies (Helbing), and stability of ecosystems of biological species or
financial firms (Cuesta).
The meeting schedule included ample time for discussion about all these issues and it was
closed by a exchange of ideas on methodology, problems of interest and hot topics, and related
issues in the field moderated by Peter Richmond, coordinator of the Action.
All participants felt that it was important to facilitate, in the manner of the physicist, close
contact between the theorist focussed on models and the researcher with access to empirical
data. In this way an interative process where theory and empirical data collection could proceed
and be developed almost in parallel with each side stimulating the other. To this end, the view
was expressed that closer contact with evolutionary biologists as well as physicists might
benefit researchers in the social sciences.
An interesting discussion around the central topic of the action,ie, competition and conflict, led
to the suggestion that comparison of different models may be a way to learn about key factors
governing those phenomena. It could therefore be useful to develop a classification of
GRUPO INTERDISCIPLINAR DE SISTEMAS COMPLEJOS
DEPARTAMENTO DE
MATEMÁTICAS
UNIVERSIDAD CARLOS III DE MADRID
Avda. Universidad, 30 – 28911 Leganés, Madrid
Tfno. 916 249 411 Fax 916 249 129 [email protected]
http://gisc.uc3m.es/~anxo
mechanisms for conflict and cooperation. These could range from resolution of political
conflicts (eg Northern ireland and the Middle East throught to a simple queuing processes. In
this respect, some sociological data such as that for friendship networks may not relate to true
evolution. However other research linked to industry such as mergers and aquisitions or venture
capital networks may prove much more relevant to this issue (e.g., about industries or
organizational networks, such as venture capital firms). Nevertheless, we recognize that even
non evolutionary data may be useful to pinpoint underlying mechanisms. A key issue is that of
understanding how the evolution of structural morphology via complex networks also
influences the dynamics. In this way one might hope to be able to predict in more detail for
example, the nature of economic booms and recessions.
The discussion concluded with a discussion of the potential benefits, in terms of increased
research funds in the maaner of the human genome and climate change, that could accrue from
coordination of multidisciplinary research across the EU.
The broad discussion of ideas of evolution from the perspective of economics and anthropology
will be discussed at a further meeting in Durham in April where we plan to consider how far
can the particle model used widely by physicists may be used to illuminate
anthropological questions together with the issue of how models may be validated
especially where data is usually sparse and noisy?
Participants have been invited to submit relevant papers and slide presentations and
subject to their agreement they will be placed on the action web site.
Below is a list of participants with extended summaries of talks and discussion topics.
GRUPO INTERDISCIPLINAR DE SISTEMAS COMPLEJOS
DEPARTAMENTO DE
MATEMÁTICAS
UNIVERSIDAD CARLOS III DE MADRID
Avda. Universidad, 30 – 28911 Leganés, Madrid
Tfno. 916 249 411 Fax 916 249 129 [email protected]
http://gisc.uc3m.es/~anxo
COST-supported participants:
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
Galam, Serge (FR)
Helbing, Dirk (CH)
Jensen, Henrik (UK)
Lambiotte, Renaud (UK)
Moreno, Yamir (ES)
Richmond, Peter (IE)
Sánchez, Anxo (ES)
Steglich, Christian (NL)
Traulsen, Arne (DE)
Vitanov, Nikolay (BG)
Non-COST-supported participants:
1.
2.
3.
4.
5.
6.
Capitán, José Angel
Cuesta, José A.
Fosco, Constanza
Grujic, Jelena
Manrubia, Susanna
Roca, Carlos P.
Topical summary (in alphabetical order of leader)
José A. Cuesta
Grupo Interdisciplinar de Sistemas Complejos
Universidad Carlos III de Madrid, Spain
Evolving robust ecosystems
Although common ecologists' knowledge has been that the larger the biodiversity of an
ecosystems the more stable it is, a result of Robert May in the beginning of the 70s pointed
in the oposite direction. May studied the stability of randomly generated ecosystems
and found that the larger the biodiversity the smaller the set of the stable ones. For a
decade experimental data seemed to confirm this result. However, the improvement
of these data and the appearance of assembly models threw away May's result indicating
that the selection occurring while building the ecosystem leads to the assembly of a
stable community resistant to invasions. In this work we propose a simplified model
of an ecosystems that permits us to study the whole assembly process. Such a process
generates a finite Markov chain containing all ecosystems connected through invasion.
Studying the chain we prove that indeed the assembly process makes the ecosystem
more and more resistant to invasions and drives it to an asymptotic robust state.
Moreover, such an state is independent on the ecosystem history. In this talk I will
describe the model and discuss its properties as well as possible extensions and the
insight they will allow us to gain.
Constanza Fosco
Grupo Interdisciplinar de Sistemas Complejos
Universidad Carlos III de Madrid, Spain
GRUPO INTERDISCIPLINAR DE SISTEMAS COMPLEJOS
DEPARTAMENTO DE
MATEMÁTICAS
UNIVERSIDAD CARLOS III DE MADRID
Avda. Universidad, 30 – 28911 Leganés, Madrid
Tfno. 916 249 411 Fax 916 249 129 [email protected]
http://gisc.uc3m.es/~anxo
On Economists and Coevolutionary Networks
"What" and "how" do economists study in this field?
As an introduction, I'll briefly present the main problems (models) and the methodological tools
used to solve them, aiming to show the differences with physicists' approach.
Next, I'll illustrate the approach through a simple model based on "Cooperation through
Imitation and Exclusion in Networks" (2008), a joint work with F. Mengel.
Finally, I'll try to outline some open research questions from the perspective of an economist.
Serge Galam
Centre de Recherche en Epistémologie Appliquée (CREA)
Ecole Polytechnique, Paris
The unexpected local spatial organization role in the emergence of new species
A simple model for speciation dynamics is presented using cellular automata. Individuals diffuse
via a synchronous random walk on a two-dimensional square lattice. Individuals can turn into a
new species at a very low rate. During each diffusive step, local fights may occur between
individuals from current and new species. Associated outcomes depend on some biased local
rules, which are independent of the overall new species density. The model’s unique ingredients
are the frequency of local fights and the bias amplitude. Each isolated new species individual is
eventually destroyed. The spreading of the new emerging species is found to obey a threshold
dynamics. It means that as long as its density is below some critical value, the current species’
individuals are rather successful in preventing any significant growth of the new comers. Even if
this threshold may be very low, it makes the stabilization of a new species totally improbable.
There exists an ongoing pressure from the new species but have no effect on the current species
domination.
However, we discover, that within our frame, below the critical threshold a new species
establishment, even at a zero measure, some very rare geometry may turn successful. Spatial
organization is thus shown to be a crucial ingredient in the making of invader killer clusters.
Once they appear by chance, they lead with non-zero probability to the total invasion of the
system. Indeed, an initial two-individual cluster is found to have a nonzero probability to spread
over the whole system. The associated phase diagram for survival or death is obtained as a
function of both the rate of fight and the bias distribution. Although the occurrence of a killing
cluster is a very rare event, it turns out to happen almost systematically over long periods of
time. Thus, after some age, survival of a species becomes random.
References
S. Galam and J. P. Radomski, Cancerous tumor: The high frequency of a rare event, Phys. Rev.
E 63, 051907 (2001)
S. Galam, B. Chopard and M. Droz, Killer geometries in competing species dynamics, Physica A
256, 263 (2002)
Dirk Helbing
Chair of Sociology, In particular of Modelling and Simulation
ETH Zürich, Switzerland
Coevolution of Cooperation and Spatial Organization in Migration Games
We study a model of mobile agents with strategic, game-theoretical interactions.
GRUPO INTERDISCIPLINAR DE SISTEMAS COMPLEJOS
DEPARTAMENTO DE
MATEMÁTICAS
UNIVERSIDAD CARLOS III DE MADRID
Avda. Universidad, 30 – 28911 Leganés, Madrid
Tfno. 916 249 411 Fax 916 249 129 [email protected]
http://gisc.uc3m.es/~anxo
Agents play a prisoner's dilemma (or snowdrift) game, move in a success-driven way and imitate
the behavior of more successful individuals. Based on this, the resulting patterns and dynamics
are dramatically different from conventional spatial games, and the level of cooperation is
increased significantly. We study the underlying success principle of this cooperation-promoting
mechanism and also its robustness with respect to invasion attempts of defectors and different
kinds of noise.
Henrik Jeldtoft Jensen
Department of Mathematics and Institute for Mathematical Sciences
Imperial Colleges London,UK
How interaction between co-evolving agents shape temporal mode and structure of the
evolving interaction network
Understanding systems level behaviour of many complex interacting agents is very challenging
for various reasons: the interacting components can lead to hierarchical structures with different
causations at different levels. We use the Tangled Nature model to discuss the co-evolutionary
aspects connecting the microscopic level of the individual to the macroscopic systems level. At
the microscopic level the individual agent may undergo evolutionary changes due to “mutations
of strategies”. The micro-dynamics always run at a constant rate. Nevertheless, the systems level
dynamics exhibit a completely different type of mode characterised by intermittent abrupt
dynamics where major upheavals keep throwing the system between meta-stable configurations.
These dramatic transitions are described by a log-Poisson time statistics. The long time effect is
a collectively adapted network.
We discuss how the systems level adaptive intermittent search is related to an increase in the
mutual information content describing the core of the population, while, at the same time, the
adaptive search is conducted through an overall network of agents described by a decreasing
degree of correlation measured in terms of mutual information. We further more relate the
systems level adaptation to the functional properties of the microscopic duplication probability.
Susanna Manrubia
Centro de Astrobiología
Torrejón de Ardoz, Madrid, Spain
Human languages and biodiversity -- Beyond the analogy
There are a good number of suggestive analogies between linguistic and biological diversity.
Mutations or linguistic change bring novelty into the system and fixation (under selection or not)
leads to divergence between species or languages. Segregating mechanisms (sexual/social
selection or spatial separation, e.g.) play an important role in speciation. The fine-structure of
species is similar to the internal diversity of certain languages, remarkably that of creole
languages. Available maps of linguistic diversity on Earth permit to undertake quantitative
measures and detect specific similarities and differences between linguistic and biological
evolution that may affect dynamical and statistical properties of the two systems.
Yamir Moreno
Instituto de Biocomputación y Física de Sistemas Complejos (BIFI)
Zaragoza, Spain
Evolutionary Dynamics on Graphs
In this talk, we discuss recent results on evolutionary game theory on complex networks. In
particular, we first analyze the Prisoner's Dilemma and discuss how cooperation emerges in these
topologies and its dynamical organization. It is shown that scale-free networks are best suited to
GRUPO INTERDISCIPLINAR DE SISTEMAS COMPLEJOS
DEPARTAMENTO DE
MATEMÁTICAS
UNIVERSIDAD CARLOS III DE MADRID
Avda. Universidad, 30 – 28911 Leganés, Madrid
Tfno. 916 249 411 Fax 916 249 129 [email protected]
http://gisc.uc3m.es/~anxo
sustain cooperative behavior, which poses the question of where did they come from. The second
part is therefore devoted to address this latter issue, and we focus on the interplay between form
and function and its role during network formation. Specifically, we propose an evolutionary
preferential attachment model based on natural selection; its main feature being that the capacity
of a node to attract new links depends on a dynamical variable governed in turn by the node
interactions. The resulting networks show many topological features of real systems. We also
point out that several dynamical features are very different from those observed in static
networks. The evolutionary preferential attachment mechanism suggests an evolutionary origin
of scale-free networks and may help understand similar feedback problems in the dynamics of
complex networks.
Christian Steglich
Sociologie/ICS
Rijksuniversiteit Groningen, The Netherlands
Statistical inference for dynamic social network data: the stochastic, actor-based approach.
Networks of social interaction are not static, but evolve over time, often in mutually dependent
feedback with changeable actor characteristics. A way of modelling this is by means of a
stochastic process on the state space of possible networks × possible distributions of actor
characteristics. Fitting such models to empirical data sets can be a challenge. Under the
stochastic, actor-based approach, it is assumed that individual actors in the network are the locus
of action, and the rules they follow in their decisions are the focus of modelling. Empirical
applications to which this technique has been applied include friendship co-evolving with
substance use, or venture capital syndications co-evolving with firm performance.
Arne Traulsen
Max-Planck Institut für Evolutionsbiologie
Plön, Germany
Evolution: From population genetics to game theory
Social dynamics models the spread of successful strategies. The mathematics of this approach is
intimately related to classical population genetics. What can we learn from population genetics
when we model social dynamics? Is there something that population geneticists can learn from
models of social behavior?