swarm-modeling
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: The Swarm space (was RePast (another agent toolkit) released)


From: Stumpo VIII, Cody (C.M.)
Subject: RE: The Swarm space (was RePast (another agent toolkit) released)
Date: Wed, 2 Feb 2000 12:39:34 -0500

RS> As we all try to be as unbiased as possible: are there
RS> comparisons, speedwise, using the same model and different 
RS> languages, or different agent tools, respectively?  
RS> Would be interesting to look at Starlogo, too.


StarLogo is interesting.  I just read Resnick's book this weekend, which
cost a mere $12 (cough, Benedikt).  It emphasizes starlogo as an aid for
thinking about distributed things.  But anyway, it does seem wildly fast,
mainly because it's not doing much.  This is really great in simulations
where not much needs to be done.  For example, the termites application in
starlogo is substantially similar to the antsort application in swarm2.
Running java starlogo and antsort  (which is straight compiled obj-c) at the
same time, the termites have the chips basically collected into one pile
before the ants have made much visible progress at all.  Note the obvious
ways in which this is not a good experiment, but anyway.  Top reported CPU
was working 50% on java, 45% on X, and 5% on antsort.  (This is Linux)

More importantly, termites requires 200 lines of code (100 of which is
readable and 100 of which is hashed...does anyone know what that stuff is?),
antsort more like 600.  (maybe 75 of which are comments)  Further, most
starlogo lines are about two words long.  The ratio of characters in the
code portions is more like 10:1.  

If we in general allow it to overcome its stigma as "that thing elementary
school children use," it could be pretty valuable for a)developing simple
physical sims, and b)providing swarm core developers an example of what
easy-to-use-for-the-non-programmer really means. (Nothin' but model)

And hey, it runs on the connection machine.  How cool is that?

Cody



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


reply via email to

[Prev in Thread] Current Thread [Next in Thread]