Re: Instantiating multiple ModelSwarms

[M Lang / S Railsback]

Fred Wan wrote:
> 
> > You and I have a lot in common. I tried that for a long long time with a
> 
> Thanks, I'll take that as a compliment ... :) I'll follow your
> recommendations (although I still have this rankling feeling that the
> ExperSwarm approach should work) and report if I run into anything
> interesting (or frustrating).

Paul knows I have to jump in here because we (i.e., Steve Jackson who
does the programming for me) did get the Experiment Swarm approach to
work. We have a cool little Experiment.Setup file where you tell it what
variables for what classes you want to have what values in the different
scenarios. The Experiment Swarm then automatically generates the model
runs (which can be in batch or graphics mode, although in graphics mode
you have to be there to push the "start" button for each run). It also
automatically generates replicates (same variable values, different
random number seed) for each scenario. (This is for a fairly elaborate
fish model.)

Some of the artwork involved was (1) writing the model so objects are
instantiated and parameterized before the Experiment Swarm reaches in
(using probes) and changes their parameter values; and (2) using a
little bit of manual memory allocation to free as much memory as
possible at the end of each model run. 

Our Experiment Swarm took quite a bit of programming (ObjC), but the
concepts and some of the code should be transferable to anybody with the
ambition. I'd be happy to provide the code and some documentation. 

Steve

-- 
address@hidden
Lang, Railsback & Assoc.
250 California Ave., Arcata CA 95521
707-822-0453; Fax 822-1868

                  ==================================
   Swarm-Support is for discussion of the technical details of the day
   to day usage of Swarm.  For list administration needs (esp.
   [un]subscribing), please send a message to <address@hidden>
   with "help" in the body of the message.