Re: matlab

[Alex Lancaster]

>>>>> "K" == Kathy and Ray La Belle <address@hidden> writes:

K> Thanks for the input on matlab. I really appreciate it.  I want to
K> create a model that the macro properties are "emerged" from the
K> molecular rules. For example, water molecules only know, feel, and
K> do certain things. Somehow, this evolves into turbulance. I have
K> the liberty of making these broad conclusions because I really know
K> very little about these fields. When molecules diffuse, they really
K> don't look at every molecule in the environment and decide where
K> they need to go. They just know something about their neighbors and
K> with a little help of randomness, they move.

K> I see a multi-layered scheme where the molecules individually act a
K> certain way. The groups of molecules act another way. These layers
K> add up in different ways until I get to the machine
K> level. Molecules -> fibers -> thread -> weave of fabric ->
K> cloth. The cloth has properties that are different from the
K> threads.

I've actually implemented (in Swarm) exactly this.  It's a 2D lattice
which allows agents (which are actually strings which represent
"molecules") to diffuse about.  In other words, objects can move
about, rather than just using the discrete PDE approach used by the
Discrete2d class, where you only have real values on the lattice
changing.  It would be nice to generalize this solution and make it
available as a Swarm library, but it's hard to justify the work when
it's only for one person, so if you're interested in working on a
joint solution, please let me know.  Within certain limitations, I've
tried to make the code pretty general already, since I had this in the
back of my mind.

It's for a project I've been doing with Walter Fontana called
"AlChemy".  See the very top of:

 http://www.santafe.edu/~alex/cas.html

I intend to be making the AlChemy code available as a tarball in the
(relatively) near future under the GPL.

K> I haven't quite convinced my management to devote the resources
K> required to create agent based models. But I am pretty sure we need
K> to use an agent based modeling system of some sort. Others are
K> doing what I want to do with swarm. Because of this, I am inclined
K> to use swarm. The other side of my organization (and I am at the
K> bottom) is familar with matlab and wants us to try it first. My
K> view is to let them use matlab (they will be busy for a while) and
K> I will use swarm ;)

[...]

Matlab or Mathematica might be OK for either prototyping or really
dumb agents, I've seen v. simple 3D particle sims in Mathematica
before (creating "rules" for agents would be a nightmare in Matlab),
but for anything else you'll need a real programming language in
conjunction with Swarm (or at least something else like it).

-- 
Alex Lancaster * address@hidden * www.santafe.edu/~alex * 505 984-8800 x242
Santa Fe Institute (www.santafe.edu) & Swarm Development Group (www.swarm.org)



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