Re: FAQ questions

[Roger Burkhart]

>From Benedikt Stefansson's message, and also in answer to Frank Lin's
questions earlier this week:

> Are you thinking of including any chapters in the FAQ on Swarm  
> objects and libraries? I have been waiting for some clearer  
> indications on how to use Swarm to create recursive swarms, i.e.  
> where agents (or the behavioral unit being simulated) are themselves  
> societies of units. For example, can each individual have his own  
> space of objects and scheduling?
> 
> Examples or hints for these kinds of simulations would also help  
> those who want to use Swarm to model hierarchies and other nested  
> societies/economies (re Frank Lin's question to this list earlier  
> this week)

Yes, the use of Swarm objects will be documented, not just informally
in a FAQ, but as part of both reference and tutorial docs.  Initial
reference docs are currently in progress for the upcoming release.

Meanwhile, we're also in process of implementing at least two-level
nested swarms just to manage the display processes of an experiment
separate from the model being simulated.  This means that examples
of multiple swarms will be close at hand in the base simulation
framework.  As part of finalizing how to do this, however, we revamped
and simplified the Swarm startup procedure more than before, so if
there had been documentation before it would now be out of date.

As far as I know, the only running examples which used multi-level
swarms before were grid9.m and mousetraps.m in the grid test program
directory (not the mousetraps application), but I don't think this is
currently being distributed since we split out the apps from the
library releases.  But all the machinery for multi-level swarms is
already in place and working, at least as exercised by these earlier
tests.  And they're very much at the core of what Swarm is intended to
be, so it's important that we have what you're asking for.

Roger Burkhart