Re: [Swarm-Modelling] foundation of ABMs

[Steve Railsback]

Pablo Gomez Mourelo wrote:
> Dear all:
>
> I am an engineer very  interested in agent-based modelling. I have a 
> question for you all, related to justification/foundation of ABMs.
> I have already read some literature and it seems to me that a 
> justification of agent-based modelling has not been achieved (Volker 
> Grimm).
> One of the problems of AB-modelling is that randomness is nearly always 
> included in our simulations, so different executions turn into different 
> outcomes.
> In comparison to mathematical models , it seems to me very difficult to 
> develop a general theory (foundation) of agent based modelling. HOw do 
> we know an ABM converges to some solution? How can we describe stability 
> of an ABM?  Many modellers feel satisfied with the graphical output, but 
> mathematicians always complain about the lack or rigour beneath the 
> simulation.

Now I will throw in my thoughts, which have been stimulated by similar 
discussions on Swarm Modelling over the years and by working with the 
good Dr. Grimm.

a. When we use agent-based models, we give up the concept of 
system-level theory. That's why we use ABM, because we don't think there 
are system-level models or concepts (like stability) that are useful for 
the kind of problems we want to work on. Instead, we are interested in 
understanding the processes by which system dynamics emerge.

b. Instead, we should be concerned with theory at the *individual* 
level: what models of how individual behave and interact with each other 
and their environment are useful for reproducing emergent system phenomena?

c. We can test and "prove" individual-level theory by putting it in an 
ABM and seeing if the ABM reproduces a wide variety of observed dynamics 
at both the individual and system levels. If so, then we can be fairly 
confident that the ABM captures the essential processes that system 
dynamics emerge from.

This kind of theory has been described as "algorithmic" instead of 
analytical- we try to understand the "rules" or mechanisms from which 
different outcomes emerge under different situations. Evolution is a 
good example: we have a very good understanding of the mechanisms of 
evolution, but we still cannot predict what the outcome will be.

There are some papers discussing and illustrating this approach at our 
site: www.humboldt.edu/~ecomodel/products.htm
Look for a paper called 'Getting "results"...', and recent papers on 
habitat selection by fish. There is a review of other examples in the 
book by Dr. Grimm and me, also described on our site.

Steve Railsback

--
Lang Railsback & Assoc.
250 California Ave.
Arcata, California 95521
707 822 0453