Re: [Swarm-Modelling] (no subject)

[Marcus G. Daniels]

hilit wrote:  
> Could you write for me the steps in need to go through in building a 
> model?
If you are going to make a working simulation yourself, don't bother 
with a flowchart.  You've got to get familiar with how to accomplish 
specific tasks in your toolkit, so I suggest spending some time studying 
simple models.   You may even find a simulation that seems a lot like 
what you want to do, but I'd suggest *avoiding* complex models with 
complex features until you can read the code of a simple model.

1) Download some package and its example models

2) Play with some of the models (e.g. how they are parameterized by 
experimenting with their controls)

3) Look deeper at how simulations in this package work by examining the 
individual code for those simulations
Get a feel for how things fit together, what the idioms and conventions 
are  (The code in NetLogo for example is particularly small and 
simple.)   E.g. how does an agent move, how does an agent get created, etc.

4) With this initial familiarity and tutorial and reference material, 
and other simulation code examples, start on a new from scratch model 
that does very simple things.  E.g. a cat moving around in a box.  If 
you can avoid it don't just rely a template.   I've seen so many folks 
become superstitious about parts they don't immediately understand, and 
then complexity and confusion escalates.  Usually some example programs 
in a package are designed to be minimalistic, so its not like you are in 
a haystack of details and have to extract a needle. 

5) For me, at least, the hardest part of any new project is breaking 
static friction.
Once you have your own simulation that works, then it becomes a matter 
of elaborating and refining it.  It just takes patience.