From COSNet

Summer school Menu Home Statistics Dynamical Systems ABM and Dynamic Networks Network Analysis Machine Learning and Data Mining Visualisation Lecture Notes

Contents 1 Agent-based Modelling Sessions 1.1 Ideas to feed back 1.2 Possible Programme 1.2.1 Common Programme 1.2.2 Introductory Stream (directed) 1.2.3 Intermediate Stream (self-directed) 1.2.4 Advanced Stream (self-directed, other languages possible) 1.3 Potential Survey for Participants 1.4 Sample code 1.5 Reading List 2 Network Analysis Session(s) 2.1 Content 2.2 Software 2.3 Software links 2.4 Reading list 3 Refs 4 Slides

Agent-based Modelling Sessions

Contributors: (so far) Matthew Berryman, Simon Angus (Monash)

Back to Summer school tutorials main page

Ideas to feed back

1. A survey for participants (see below)
2. Ordering the NetLogo/Pajek tutorials before the statistical tutorial such that students can analyse the outputs of their modelling (in say, R)

Possible Programme

Assumption 1

Most users will not be familiar with NetLogo.

Assumption 2

Some users will be proficient in other languages, and/or computer programming in general and will want to move at a faster pace.

Common Programme

1. Introduction to NetLogo
   1. Software environment
   2. Programming language
   3. Simple examples
2. Discussion of 'streams'

Introductory Stream (directed)

1. Reproducing CA results (e.g. Wolfram-esque type I, II, III, IV behaviour)
2. Reproducing agent-based modelling results (e.g. programming agents to play the 'protector' 'attacker' game (have you played this in real life -- very nice example of simple rules --> complexity!)
3. Reproducing dynamic network results (e.g. preferential attachment). This then gives them some of their own data for the network analysis section.

Intermediate Stream (self-directed)

1. (optional) Work through introductory material at own pace, self-directed
2. Try out problem of own choice, with suggestions (for those who'd like them). e.g.
   1. Model a pack of riders in the Tour de France
   2. Sand in the sand-pile
   3. Sub-way tunnel traffic
   4. Social network formation (friend of a friend etc.)
3. Short presentations given periodically on:
   1. Behaviour Space
   2. Applets
   3. Limitations of NetLogo (e.g. getting data out, limited dimensionality, scaling up)
4. Include dynamic network as part of model. This then gives them some of their own data for the network analysis section.

Advanced Stream (self-directed, other languages possible)

• Is this required?
• We would need developer tools for other languages installed.

Potential Survey for Participants

To be administered approx. 2 weeks? prior to workshop to make any last minute programme alterations to the Computing component

1. Have you built an agent-based model before?
2. If so, please tick wich software you used:

• • NetLogo
  • C(+)(++)
  • Matlab
  • Java
  • Python
  • Other .....

1. If you have ticked 'yes' to "NetLogo" above, please indicate the level of proficiency you believe you hold in these areas:

• • I can use patches and turtles to make a simple interaction model
  • I can most of the following features: plots, lists, reporters
  • I have run experiments using the BehaviourSpace tool before
  • I have built an agent-model using the 'links' class before
  • I have successfully exported an applet before

(... with some kind of 'Strongly Agree' ----> 'Strongly Disagree' ,)

Sample code

Download link: Image:NetLogo code.zip

Reading List

Papers

• Macal, C. M. and North, M. J., Tutorial on Agent-Based Modeling and Simulation (part 1), Proc. 2005 Winter Simulation Conference, M. E. Kuhl, N. M. Steiger, F. B. Armstrong, and J. A. Joines, eds.
• Macal, C. M. and North, M. J. Tutorial on Agent-Based Modeling and Simulation (part 2), Proc. 2006 Winter Simulation Conference, L.F. Perrone, F. P. Wieland, J. Liu, B. G. Lawson, D. M. Nicol, and R. M. Fujimoto, eds.
• Wolfe, S. R., Sierhuis, M. and Jarvis, P. A., To BDI, or Not to BDI: Design Choices in an Agent-Based Traffic Flow Management Simulation.
• Berryman, M. J., Review of Software Platforms for Agent Based Models, DSTO-GD-0532, 2008.
• Leombrunia, R. and Richiardi, M., Why are economists sceptical about agent-based simulations? Physica A: Statistical Mechanics and its Applications; Volume 355, Issue 1, 1 September 2005, Pages 103-109.
• Holland, John H, and John H Miller. 1991. "Artificial Adaptive Agents in Economic Theory." American Economic Review, Papers and Proceedings 81:365―370.

Books

• Handbook of Computational Economics Vol 2 edited by Leigh Tesfatsion and Kenneth Judd, North Holland, 2006.
• Miller, J. H. and Page, S. E., Complex Adaptive Systems: An Introduction to Computational Models of Social Life, 2007.
• Gell-Mann, M., The Quark and the Jaguar, 2002. Murray Gell-Man is a Nobel prize winner, founding member of the Santa Fe Institute; this book has excellent advice about the 'Art' of writing good models)
• Holland, J. H. Emergence: From Chaos to Order, 1999. John Holland is the 'father' of genetic algorithm approaches, widely used as a tool in complex systems ABMs)

Network Analysis Session(s)

Contributors: (so far) Matthew Berryman, Simon Angus (Monash)

Content

• PAJEK Tutorial by Simon

Other notes:

• Pajek Tutorial (link dead, email Matthew for the .doc file) - the material in this should be extended.

Software

Suggest we stick to one package of software (Pajek?). There are pros and cons of using Pajek, and also for Matlab. On the one hand, Pajek supports point and click stuff, but that doesn't necessarily give you a good understanding of the algorithms (mitigated by the fact that we should be teaching this), and often the documentation is poor. So perhaps use Pajek and then mention other alternatives.

Software links

• Pajek
• Software for random hierarchical graphs
• Pajek file format (also supported by JUNG)
• JUNG - Java Universal Network / Graph framework
  • JUNG jar file
  • JUNG documentation
• To use JUNG you will also need:
  • COLT jar file
  • Xerces (only if you want to read Graph ML files).
• Eclipse Java development environment for Windows
• Tutorial on using Eclipse to start a new Java project
  • Note that this window has the option for External JAR files (which you can use to include the COLT and Xerces files).

Reading list

Papers

• Barabasi, A.-L., and Albert, R. "Emergence of scaling in random networks". Science 286 (oct 1999), 509–512.
• Watts, D. J., and Strogatz, S. H. "Collective dynamics of ‘small-world’ networks". Nature 393 (June 1998), 440–442.
• Girvan, M., and Newman, M. E. J. Community structure in social and biological networks, Tech. rep., Santa Fe Institute and Cornell University, 2001.
• Batagelj, V. and Mrvar, A. Pajek: Program for Large Network Analysis
• Clauset, A., Newman, M. E. J. and Moore, C., Finding community structure in very large networks, Phys. Rev. E 70, 066111 (2004).
  • See also Aaron's blog post

Books

• Watts, D. J., Six Degrees: The Science of a Connected Age, 2003.
• Barabási, A.-L., Linked: How Everything Is Connected to Everything Else and What It Means, 2003.

Related

• Clauset, A., Shalizi, C. R., and Newman, M. E. J. Power-law distributions in empirical data, arXiv pre-print, 2007. (How-to on fitting power-laws to your data, including the dos and don'ts, and goodness of fit tests)
  • See also So you think you have a power-law — Well Isn't That Special?

Refs

• ANU Physics Summer Schools page

Slides

Link