Re: Schedule Properties

[Nelson Minar]

Good questions, Manor, which I'll boil down to "Swarm supports all
this really complicated scheduling - but do the users need that?" I
ask myself that question too. I'm about to teach the more complicated
stuff today to my class. But only on a theoretical level, I don't have
any examples to show.

>most A-Life researchers are interested in emergent phenomena they
>seem to always seek the basic pulse of the system they are modelling,
>and they seem to prefer getting the rest by pure observation (i.e.
>the higher-level, slower events are observed not scheduled). Also,
>most systems seem to have a multitude of agents which require
>updating at _that_ basic rate.

I agree that most Alife simulations have this basic form of schedule.
Time-step simulations. I think it'd be a good thing to support that
style of simulation more simply, maybe with a subclass of some sort.

But it's important to keep the more complicated schedule possibilities
in Swarm and to leave them as the fundamental building blocks. Why?
Because time-step models, while common, aren't good enough. Swarm in
part servers to prod the simulation community into developing more
sophisticated modelling. Sophisticated scheduling is part of that.

The real problem is there aren't many examples of models in Swarm that
use fancy schedules. Maybe not really any - there's heatbugs and
mousetrap, is anything else available in source form really that
different? I think a priority for the Swarm project is to somehow get
a model out there that uses all the fancy features of Swarm, to show
what they're useful for.

There's another reason to leave in fancy scheduling, particularly
subswarms. It's fundamentally cool. It's new, too, the one unique
thing about the Swarm system. These features deserve to have a life,
to be demonstrated, to be written about.

More work for the Swarm team. Easy for us to say now that we're gone :-)


                  ==================================
   Swarm-Modelling is for discussion of Simulation and Modelling techniques
   esp. using Swarm.  For list administration needs (esp. [un]subscribing),
   please send a message to <address@hidden> with "help" in the
   body of the message.
                  ==================================