Re: [Swarm-Modelling] SWARM on Clusters and "Models in the Wild"

[Steve Railsback]

Darren Schreiber wrote:

> ...

> In general, it isn't very easy to figure out what other people are doing 
> with their modeling or with their testing.  In my little fantasy world, 
> I imagine reading an article and then seeing a link to their "Evaluation 
> Summary" webpage that contains copies of their code along with a variety 
> of tests and results.  Since this is all very computationally intensive, 
> I'd want the Evaluation tool to be able to send the model off to 
> different processors (via a grid, ftp/ssh, cluster, or whatever) and get 
> back the results.
>
> None of what I have described requires miracles of programming.  But, it 
> would certainly require some coordinated effort, vision, and probably 
> funding.  I do think it would help me sleep better at night in believing 
> my modeling and the modeling of others.  And, I believe it would provide 
> the kind of "ABM advocacy" that no amount of persuasive theorizing could 
> produce.

Hi Darren,

We are just now sending a couple of innocent young math grad students 
off on the adventure of trying to figure out how to do such analyses of 
a big, complex IBM of fish. (At a little state university that is, 
unfortunately, one of the few remaining schools that does not have the 
world's 3rd fastest computer [cluster].)

There are many dimensions of input to look at model sensitivity to: 
parameters, input data, the environment you place the individuals in, 
structural assumptions of the model.

The only promising potful of literature we've found so far is work by 
Volker Grimm's colleague Martin Drechsler on analysis of other (not 
agent-based) big models for how uncertainty affects decisions based on 
the models: http://pinus.oesa.ufz.de/martind/publikationen.shtml

If anyone knows of any other literature, please say so! (Our student did 
a Google search on "latin hypercube analysis" and the top response was 
the email I sent to Swarm Modelling asking if anyone had implemented 
latin hypercubes in swarm.)

Still- you can good science with an ABM before doing all this detailed 
analysis. For example, by comparing alternative rules for agent behavior 
and seeing which produce wide varieties of realistic patterns of system 
behaviors. If your system behavior patterns are general, but you look at 
many of them, you can draw strong and robust conclusions about agent 
behavior without having to worry much about model sensitivities & 
uncertainties.

Steve
--
Lang, Railsback & Associates
250 California Ave.
Arcata CA 95521
707-822-0453; fax 822-1868
address@hidden