Re: [Swarm-Modelling] beginner

[Marcus G. Daniels]

Hilit,
> I wish to explore stray cat dynamics in my city by using an individual
> based spatially explicit model.
I think you may find that it is easy to spend a lot of time on making 
realistic terrain and describing lots of details in how your cats 
interact with the terrain.   (And potentially not really be sure that 
these details are true, esp. with regard to behaviors.)  Unless you have 
strong reason to think that these things will matter for overall 
properties of stray cats, you may find it useful to start with a very 
abstract description of the city using a package like NetLogo in order 
to develop some zeroth order ideas about how the dynamics work.

Supposing that sooner or later you have strong opinions on what matters 
and what doesn't, then you may require more realism and agent complexity 
and thus grow a substantial and specific wish list for a simulation 
implementation.  When you find that the absence of these features are 
blocking your ability to ask questions and get answers, then you've 
moved into more of a simulation engineering mode.   Here I'd certainly 
echo Paul's advice by suggesting then would be a good time to either 
find a programmer to work with, or else learn some programming.   
Further, to do so by learning fundamentals first (using programming 
tools in a relatively `raw' way so that you understand them) and then 
moving on to fancy IDEs that in some sense encapsulate but also hide the 
fundamentals.    Further, once you are comfortable programming, you'll 
probably just do everything using a programming approach, as it is fully 
general.  (Potentially even without any toolkit.)

Ultimately, if you want to use your model to do predictions, then you'll 
be limited by data and data processing capability, and your ability to 
explore the parameter space with your simulation(s).  Here's another 
reason for learning programming and system administration.  You'll have 
plenty of long running jobs and be wanting to run them over multiple 
computers for different parameter configurations.

Marcus