Re: [Swarm-Support] Axelrod models

[Paul E. Johnson]

There are many different Swarm versions of the culture model I know of. 
I am not sure which one you refer to below, but I do know that Lars Erik 
Cederman (who was at UCLA) did one for his class that was nice. Even 
before that,  Rick Riolo (at U Mich) did a very nice, bigger one.  Lars 
Erik's model had the appeal of clarity in design, while Rick's had some 
nicer presentation features.

In about 1998 I started working on many variants of that model and 
built a quite large Swarm model on opinion networks.  An intermediate 
version that probably has the most resemblance to the original Axelrod 
model is this one:

http://lark.cc.ku.edu/~pauljohn/Swarm/MySwarmCode/AxelrodCultureModelPlus/cultureWrappedListBatchRaster-990823.tar.gz

That's the one that is described in this paper:
http://lark.cc.ku.edu/~pauljohn/ResearchPapers/APSA99/
As you will see from that paper, I was still quite seriously 
trauamatized by the idea that I might need to write a whole new 
container class.  But I got over it.

That model got retooled quite a bit to build in just about every bell 
and whistle I could find, including a command line argument processor, 
full serialization, and also 3 different decentralized scheduling 
approaches.  That's the one that is discussed in a book that Bob 
Huckfeldt, John Sprague, and I have written and is being processed at 
Cambridge University Press.  Last time I heard, it had a title 
"Durability of Disagreement" but there was talk of change.   I am not 
allowed to give out copies of that now, but I see I still have one 
convention paper about it here:

http://lark.cc.ku.edu/~pauljohn/ResearchPapers/MidWest02/mw02.pdf

I have the source code for a late version of that code here, and if you 
look in here :

http://lark.cc.ku.edu/~pauljohn/Swarm/MySwarmCode/OpinionFormation

Look in Opinion-Docs and you should see some docs in HTML format.  In 
the code for the Opinion model, there is some markup and I recently 
converted it so it could be post processed by a program called 
autogsdoc, which is part of the GNUstep base. Its not the best 
documentation in the world, but is the best that was within my ability.

In the comments about the ModelSwarm class, there is this one thing I 
want to point out to you: The parameter "ModelType" controls how agents 
are put together for interaction and what they do in response to 
interaction.  The Axelrod approach is ModelType 0.

(id) *createCitizens*;

This method does the heavy lifting of creating the agents that are put 
into the citizenList. The ModelType variable (see the Parameters object) 
determines which sort of agents are created. These are the types:

   * ModelType 0: Axelrod (for both select and adjust)
   * ModelType 1: HJ type citizens (for both select and adjust)
   * ModelType 2: Axelrod select, HJ adjust, persuasionModel=0
   * ModelType 3: Coleman type agents (Coleman select, Axelrod adjust)
   * ModelType 4: Coleman type agents (Coleman select, Axelrod adjust).
     Same asModelType 3 in current implementation.
   * ModelType 5: MikePM2 type agents
   * ModelType 6: Axelrod Select, MikePM2 adjust
   * ModelType 7: HJ Select, Axelrod adjust
   * ModelType 8: Axelrod select, HJ adjust, persuasionModel=1




Good luck!

pj

steven brandenburger wrote:

> Hi
>  
> I just procured a copy of Complexity of cooperation by Axelrod. Its an 
> amazing read. I wanted to play around and see the results of the 
> models described. The book points to its web site for some of the 
> models. But these are in PASCAL !!!!!. I did find the swarm version of 
> the culture model at Davidson's site. Aren't there any other swarm 
> versions of the models described.... the IPD (using genetic algos) and 
> the Tribute model??
>  
> I would be grateful if somebody on these lists could give me some pointers
>  
> regards
> steve
> ------------------------------------------------------------------------
> Do you Yahoo!?
> Exclusive Video Premiere - Britney Spears 
> <http://launch.yahoo.com/video/?1093432&fs=1&redirectURL=http://launch.yahoo.com/promos/britneyspears/> 




--
Paul E. Johnson                       email: address@hidden
Dept. of Political Science            http://lark.cc.ukans.edu/~pauljohn
University of Kansas                  Office: (785) 864-9086
Lawrence, Kansas 66045                FAX: (785) 864-5700