Re: [Swarm-Modelling] lifecycle requirements

[Scott Christley]

On Nov 25, 2006, at 6:26 AM, glen e. p. ropella wrote:

> My question to the list is:  What, if any, modeling level requirements
> do you think are important and does the outline I provide above  
> target it?

A major requirement to me is how models are to be specified.  And I  
don't mean what language, programming or otherwise, that it is  
encoded in.  Should models be specified declaratively or  
imperatively?  (Glen, I'm not sure how this relates to your synthetic  
vs inductive)  Declarative means you say "what" while imperative says  
"how".  Most programming is imperative, people who are experts at  
making computers do things know the best way "how" to make them do  
things.  Because simulation is programming task, it is easy for the  
modeling to get overwhelmed by the "how" of programming.  However, I  
don't think this is the best way to do modeling.

My requirement is that you never specify a single model, you specify  
a space of models.  And you do this by declaratively specify "what"  
you want the model to produce through a set of observations,  
experiments, assumptions, axioms, etc.  All of these declarations  
work together to constrain the infinite space of models down to a  
finite set of models, these models then become hypotheses which you  
test through simulation i.e. you run simulations for each model to  
produce outcomes.  Sometimes you can throw out additional models  
because they produce unrealistic outcomes, while the realistic one  
now provide experiments which can (hopefully) be performed against  
the real phenomena.


Scott