Re: [Swarm-Modelling] Re: [Swarm-Support] Re: Agent communication and grid models

[Jason Alexander]

On Tuesday, May 27, 2003, at 07:16 Europe/London, Kanagaraj Krishna 
wrote:

> Can i know in detail what is meant by a "soup" model.

I don't think this is a term of art; if it is, I wish it would be 
renamed.

In economics, one common type of model used by evolutionary game 
theorists is called a "local interaction model" (for a dated reference, 
see Ellison, "Learning, Local Interaction, and Coordination", 
_Econometrica_, 1993). In this type of model, interactions among agents 
can only occur along the lines of a predefined network.  This is also 
the kind of model Watts had in mind in _Small Worlds_ when he talked 
about games played on graphs.  I think this is what Darren meant when 
he was talking about a "topological" solution (although the 
restrictions on interactions conveyed by the underlying network need 
not correspond to any natural topology).

A second model for evolutionary games is a random mixing model - the 
kind of model which underlies the replicator dynamics. In this kind of 
model, no constraints on possible interactions exist, so everyone can 
interact with everyone else.  One often uses randomization to select a 
subset of individuals from a population to interact with, or just has 
everyone interact with everyone else in a round-robin kind of 
tournament.

Then there are the means between these endpoints which aren't easily 
classified. Like the model of social network formation by Skyrms and 
Pemantle (Proceedings of the National Academy of Science, August 1, 
2000, vol. 97, no. 16), for example.

Cheers,

Jason
--
J. McKenzie Alexander
Department of Philosophy, Logic and Scientific Method
London School of Economics and Political Science
Houghton Street, London WC2A 2AE