[ Initiative ][ Introduction ][ Overview ][ Supported Swarm versions ][ Compiler Download Page ][ Compiler Usage Guide ][ Tutorial ][ Course Outline ][ Examples ][ Reference Manual ][ Papers ][ Comments for Swarmites ][ Technical Manual ] ][ About ]

Social Science Computer Models

[Case Studies, Tutorial Examples]
[Ongoing Projects]
[Experiments in Stigmergy]

This page gives a very quick overview of the computer models developed and under development at Agent-lab. The detailed descriptions of these models can be found elsewhere on the web or in printed form. We have an incentive to make them all available on the World Wide Web in electronic form, but till that time, please contact us directly in case of any difficulties.

Case Studies, Tutorial Examples

These models were developed mostly for educational purposes and serve as examples in our modeling courses. They are replications of previously published models. As these original models are usually not individual (or agen) based ones, the replication also meant the redesign of the subject model. The transformation of an ordinary model into an agent based one is a pretty interesting and up to now not well studied problem, so these experiences also gave us very important insights into the fundamentals of agent based modeling, and modeling in general. Some of these models also served as case studies for the development of our modeling environment.

Introductiory Experiments with Logistic Lattice Maps

These were the first experiments carried out in the Systems Laboratory. They were used to design the fundamentals of our modeling environment

Click on the thumbnail on the rigth to see the full picture.

Replication of Peter Allen's Population Dynamics Model

This very interesting model of population dinamics served ( click here for the description of the model) as a good lesson to learn and to teach the basics of computer modeling and simualtion. We also used this model the study the possible visualization methods of computer simulation results. Follow this link for more details on this study.

The original implementation is shown on the picture to the right. Some other method of visualization of the same result set is shown below. 

Cllick on the thumbnails to get the full pictures.

Replication of Paul Krugman's Edge City Model

This minimal model of self-organization of urban places was published as a demonstration in Paul Krugman's book on The Self-Organizing Economy. As the original model wasn't really agent-based, we reimplemented it using this novel approach. This project gave us very deep insights into the hidden assumptions of the original model, and into the problems which occur when the way of thinking shifts from top-down to botton-up, which is one of the main challenges of agent (or individual) based modeling and simulation.

The transformation project resulted in various results, one of the most important of which is a case study we' ve developed. We use this material in our education to give insights into the fundamentals of this novel approach in modeling. The final version of the replicated model also serves as an example of our MAML language.

The details of this project are discussed elsewhere.

On the right a series of pictures are presented showing the process as central places evolve. (Click on the thumbnails to get the full pictures.) For more explanations and more results, please refer to the separate page mentioned above.



Replication of Thomas Schelling's Segregation Model

This fairly simple but excellent piece of work is somewhere in the roots of social science computer simulations. Moreover, although it was developed decades ago, it fits pretty well into the concept of agent based modeling.

Thanks to its simplicity this model can be used as an introductory example at our Social Science Simulations course, and as a programming example of our MAML language.

These are screen captures of the simulation. Click on the thumbnails to get the full pictures.

Ongoing Projects

The following models are under development, so only a limited set of results is available. In this section we give a short overview of each projects including as much illustration as we can without taking serious resources from the real development. The results of these projects will appear on-line as soon as possible.

Dynamical Economy: The Case Of Inflation (UNDER DEVELOPMENT)

This project is the last one in the series of our studies examining the transition from traditionally implemented models into agent based ones. The replication will serve as an example of the modeling environment we have developed. The model may also be subject of further improvements and studies.

The original model was written by  Perazzo, Reich, Schvarzer, and Virasoro, and was published in Complexity, col 2, No 6, 1997.

A Simple Economy Model (UNDER DEVELOPMENT)

This project is aiming at the replication and further development of Alex Viskovatoff's Model of Simple Economy.

The Senegal Model: A Modified Version (UNDER DEVELOPMENT)

This projects is about modeling a country's (let's say Senegal's) population, economy and politics (to some extent). It was highly inspired by Peter Allen's work on modeling Senegal.

A Model of Fast Food Restaurant Chains (UNDER DEVELOPMENT)

In this model of urban development different strategies of unit location for competing fast food restaurant chains are explored based on real GIS data of Budapest.

On the right and below a couple of pictures are presented illustrating the kind of work we are doing. (Click on the thumbnails the get the full pictures!)

The roadmap of Budapest showing the means how different locations in the town can be accessed. Three reastaurant units (of different chains) are also indicated.

The cost their consumers need to spend on travelling, calculated by two different restaurant units, based on the roadmap above.

The influence of the indicated restaurant units on costumers.

A Simple Model of Common Decision (UNDER DEVELOPMENT)

The main objective of this project is to give another example of building models in MAML. However, it also tries to capture some interesting behaviour of human societies, namely the process of making common decisions effecting the community. In the future this oversimplified model might be subject of further development.

Experiments in Stigmergy

The concept of stigmergy was introduced by a French biologist, P.P. Grassé in the '50s when he was studying the behaviour of social insects. The word itself comes from two Greek words: stigma (outstanding sign), and ergon (work) giving the meaning incentive to work by the product of work itself. The concept vaguely is a coordination mechanism, which allows fairly simple aminals to achieve highly ordered emergent behavior at the level of the society, composed of simple activities through indirect communication. One of the most exciting things about stigmergy is that it leads to pretty robust emergent phenomena, which remain virtually unchanged under heavily changing circumstances.

The concept gained new acceptance in the field of Artificial Life in the '90s, and in Computer Sciences very recently. Although the origins of stigmergy are pretty much connected to social activity and behaviour, little if any work were devoted to study the mechanism in the context of human societies, while it seems to be a natural concept to describe some phenomena in our everyday life. Think of the way how culture (e.g. litterature or science) evolves: to some extent this common product of human kind can be seen as a product of loosely coordinated activities of millions of agents, each of them being inspired by the perceived parts of the common product (the product of the others). Or think of the self-organizing structure of the Internet, especially how new techniques, standards and concepts spread. These all could be modeled under the general concept of stigmergy.

If you are interested in some interesting ideas of Mitchel Resnick on the decentralized world around us and our centralized mindset, follow this link.

Deneubourg's Path Planning Ants

This model is a famous one, often cited in the context of stigmergy. It nicely demonstrates that individuals in a society can produce highly coordinated behavior at the level of the society, even if they are not aware of themselves doing that. 

The model captures the ants food seeking behaviour. Ants are wandering around in the space looking for food while leaving trails of pheromone (chemical sign) behind. They tend to follow beacon trails left by homing ants carrying food. When they eventually find the food source, they try to get home by tending to follow the pheromone path and leaving beacon trail behind. The pheromone and the beacon evaporates over time.

This model was developed in part to study the technical details needed to develop our simulation environment, but it was also very useful in understanding the concept of stigmergy.

On the right there is a series of screenshots presented from one typical run of the simulation. (Click on the thumbnails to get the full pictures!) The food source is represented by the blue dots in the upper left corner of the space. The nest is in the middle of the grid, denoted by a yellow dot (which is often hidden by the ants). Ants looking for food are the white, while ants carrying food are the green dots on the screen. The screeanshots are divided into too parts, one of them showing the concentrate of pheromone, another showing that of the baecon.

The series shows as the ants start out from the nest (Figure 1), wander around (Figure 2), begin to build the trail (Figure 3 and 4), and tend to stabilize the trail and remain on it. Note however, that this run was not perfect in the sense that a number of ants remained off the path built by the others.

Figure 1
Figure 2
Figure 3
Figure 4
Figure 5

Bonabeau's Model on Cemetery Organization in Ant Societies

This is one of the core models of the concept of stigmergy, and in broader sense that of decentralized self-organizing systems. It is fairly similar to Deneubourg's Model of Breed Sorting Ants.

In this model agents acting upon local information (from their environment) collect entities (breed or dead bodies) initially scattered around the space into clusters, without any explicit coordination or communication amongst each other

The model was used to gain deeper understanding of the concept and also to study the underlying coordination behavior. The illustrations to the right come from a slightly modified version of the original algorithm, which was developed as an example of our MAML language. Please find the detailed description there.

Click on the thumbnails to get the full pictures!

Uniformization of Local Culture (UNDER DEVELOPMENT)

This model was inspired by Deneubourg's Breed Sorting Ant Society Model and tries to capture the process how different cultures carried by moving agents tend to result in spots with uniform local culture.

The model is still being developed, but some screenshots are available on the right to illustrate it. The green and white dots are the agents wandering around, while the red dots represent the level of uniformness of the culture at the given location. The brighter the red is, the more uniform the local culture became.

Click on the thumbnails for the full pictures!


copyright © 1999 agent-lab
Send comments to maml@maml.hu