Re: Instantiating multiple ModelSwarms

[SCHREIBER,DARREN MATTHEW]

As a plug for Paul, I have used his methods from the ReapeatingHeatbugs
for both my housing segregation and my party formation models.  I didn't
want to go through the time of learning drone last summer but I needed to
do multiple runs.  It worked really well for me.

I actually left the graphics on since I wanted to look at end states from
the runs of the models.  I also included some code to have the model save
the last images from each run.  And, I had the code spit out the data from
the model as text so that I could easily import it into Excel for
analysis.

So this is both an endorsement and a thanks for Paul's work in developing
useful Swarm code that has been a great help to me.

        Darren


On Tue, 24 Jul 2001, Paul E Johnson wrote:

> OK, I can't be modest anymore.  I hate to see you reinventing the
> wheel,  my code has been here all along:
> http://lark.cc.ukans.edu/~pauljohn/Swarm/MySwarmCode/RepeatingHeatbugsParameters-1.4.1-4.tar.gz
> 
> That is from last year, and to compile you need make
> EXTRAOBJCFLAGS=-Wno-error because there are unused variables.  But I've
> just tested it just now, it runs.
> 
> It does what Fred wants. It starts a control panel, click start or next,
> it pops up an experiment, the second control panel can start and stop
> the simulation, and when it runs to completion, you can click next to
> make another experiment/control panel start up.  
> 
> Here's the problem. If you don't let each individual experiment run to
> completion, and instead you hit quit inside the experiment control
> panel, the whole thing crashes.  
> 
> $ ./heatbugs
> *** event raised for error: InvalidOperation
> *** function: _update_mergeSchedules(), file:
> /home/pauljohn/LinuxDownloads/redhat/BUILD/swarm-2.1.75.20010613/src/activity/Schedule.m,
> line: 276
> > MergeSchedule is invalid. There is no
> > mergeAction for schedule where action
> > should be inserted!
> *** execution terminating due to error
> /home/pauljohn/LinuxDownloads/redhat/BUILD/swarm-2.1.75.20010613/src/defobj/Symbol.m:173
> -[Error(c) _raiseEvent:]
> Aborted (core dumped)
> 
> I THINK there was a brief period of time when it did not crash with
> swarm in the 1.4x era, but I can't promise anything, because, at one
> point, I asked Marcus about it, and he said "yes, that crashes."  He
> looked at the code once or twice and said, "yep, that doesn't work." 
> And I've not looked at it since.
> 
> I spent quite a long time on that. I thought it should become the
> prototype of all gui models, because it works like a "reset" button and
> you don't have to restart a simulation.  I was thinking it was a good
> use of effort and a potential prototype for repeated simulations, but
> then Marcus and I talked for a long time and his advice then was the
> same as now: it is generally easier and more foolproof to replicate
> experiments by running over and over with command line parameters and
> this experiment swarm-type-thing is needed only if one run learns from
> the calculations of the previous one and you don't want to accomplish
> the same learning by serilization.  
> 
> I was doing that because I could not figure how to make Drone work, but
> then I figured that out with the help of our good friends in Michigan,
> and now I have written replicator.pl, a lightweight perl approach to do
> the same thing.
> 
> "Marcus G. Daniels" wrote:
> > 
> > >>>>> "FW" == Fred Wan <address@hidden> writes:
> > 
> > FW> The problem is that I _do_ want two control panels. One to start a
> > FW> ModelSwarm and another to stop the ModelSwarm while it is
> > FW> running.
> > 
> > I actually don't know what Paul is talking about.  If you run into a
> > specific bug where you can say "this is definitely wrong", then I'll
> > fix it.  Or, if you have a specific behavior that confuses you then I'll
> > try to give you an answer.
> > 
> > Incidentally, an updateDisplay method is now in SwarmEnvironment.
> > 
> >                   ==================================
> >    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.
> 
> -- 
> Paul E. Johnson                       email: address@hidden
> Dept. of Political Science            http://lark.cc.ukans.edu/~pauljohn
> University of Kansas                  Office: (785) 864-9086
> Lawrence, Kansas 66045                FAX: (785) 864-5700
> 
>                   ==================================
>    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.
> 
> 


                  ==================================
   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.