Re: FW From Simsoc: RE: Sugarscape - the landscape formula

[Nelson Minar]

(feel free to forward this reply to SimSOC)

>I saw this note on the SimSOC list, and there were no answers,
>so I'm hitting the Swarm group.
>What formula is used to generate the sugar distribution in Sugarscape?

In the Swarm version of SugarScape that I wrote last summer ("sss", at
ftp://ftp.santafe.edu/pub/swarm/users-contrib/anarchy/sss.tar.gz), I
just hacked it up by hand. That's why it's a PGM file (a picture) and
not a formula. I don't remember now if I exactly reproduced the figure
or just got close, and it's possible I actually took a screenshot from
their CD and digitally turned that into a datafile. Anyway, it's ad hoc.

>>>Does anyone know what equation Epstein and Axtell used to generate the
>>>distribution of sugar on the sugarscape?  The formula
>>>sugar = 25 * sin (x-25)/4 * sin (y-25)/4

I believe their map was ad hoc as well. You might want to ask them
yourself. It's a weird distribution, not quite symmetric. Note that
the two piles are not centered respectively and they're quite close to
each other around the corner if you assume a toroidal topology. Also
there's a bulge between the two piles to makes the valley between them
narrower than you might expect.

This question raises a good general point about simulation - it's
impossible to exactly reproduce someone's model based on a published
description. The Sugarscape book is quite good at describing their
model (much better than most of the field!), but even so there's enough
left unexplained that you won't be able to exactly reproduce their
results. In my experience reimplementing parts of SugarScape I could
qualitatively reproduce what they report, but only with a disturbing
amount of tinkering. This is a general problem with computer modeling.

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