Newsgroups: comp.ai.alife
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!oitnews.harvard.edu!purdue!lerc.nasa.gov!magnus.acs.ohio-state.edu!math.ohio-state.edu!howland.reston.ans.net!news.sprintlink.net!simtel!swidir.switch.ch!swsbe6.switch.ch!news.unige.ch!usenet
From: BAUJARD Olivier 26272 <baujard>
Subject: SMA vs CA
Content-Type: text/plain; charset=us-ascii
Message-ID: <1995Sep20.100839.24147@news.unige.ch>
To: olb@diogenes.hcuge.ch
Sender: usenet@news.unige.ch
Content-Transfer-Encoding: 7bit
Organization: University of Geneva
Mime-Version: 1.0
Date: Wed, 20 Sep 1995 10:08:39 GMT
X-Mailer: Mozilla 1.1N (X11; I; SunOS 5.3 sun4c)
X-Url: news:comp.ai.alife
Lines: 86

Hi,

I'm new in that newsgroup and I would like your opinion concerning some work
I'm involved in.

I developped some years ago a simulation tool based on the following features.

I first build the Voronoi tesselation of a set of points (germs) that gives me
a lattice. That lattice is a set of polygons built from the germs and having
the Voronoi property: each point in a polygon is nearer from the associated
germ than from any other germ.

Values can be associated with each polygons.

To each germ (polygon) I also associate an entity (called cell) provided with
some internal rules. Those entities can be heterogeneous.

Then, after having provided cells with rules and initialized the values
associated with polygons, I run a simulation.

At each step, cells synchronously change their internal state (modification of
some variables) according to their neighborhood (state of cells associated with
the neighbor Voronoi polygons) and/or the values of variables associated with
the corresponding Voronoi polygons.

Cells can also die (supression of the corresponding germ and polygon in the
Voronoi lattice) or can give birth to a new cell (insertion of a new germ in
the Voronoi lattice).

Thus, at each step, the lattice is modified as well as the neighborhood of
cells.

Complex behaviours are exhibited that could correspond to Class IV behaviours.
Indeed, with some simulations we obtain cyclic behaviours (repetition of
patterns in the lattice) or stable patterns and a stability 

Typical applications are biological ones (growth cell simulation or forest
simulation). For more details you can have a look to the following URL:
http://expasy.hcuge.ch/sgaico/html/olb/Mosaic.html

In the growth cell application, lattice a polygon correspond to the local
environment associated with a biological cell. The cells are the biological
cells.
Those cells are provided with some simple biological rules. To summarize, a
cell will divide if a sufficient amount of some hormon is present in its local
environment or will become differentiated if a sufficient amount of an other
hormon is present in its local environment. Cells will also die. Hormons are
values associated with polygons and a diffusion law as been introduced that
tends to equilibate concentrations in polygons at each simulation step. Results
of simulation exhibit degenerated, cyclic or stable behaviours according to
initial conditions.
Those behaviors are typically biological behaviors. A very interesting
behaviour is the stability (stability in terms of cell number, in terms of
concentration in polygons and in terms of patterns).

As I come from the Distributed Artificial Community, I used to call those
simulations Multi-Agent Systems or Reactive Multi-Agent
Systems. But they also share properties with CA (more precisely inhomogeneous
CA or structurally dynamic CA).

So, here are my questions:

Can I call those simulations CA ?

To your opinion, what are the differences (common points) with CA ?

Do you know about similar or related work (if not based on Voronoi lattice, at
least on complex dynamically computed lattice) ?

Is a reactive multi-agent system a kind of CA?

Thanks a lot for you answers and comments.

Olivier

-- 
-------------------------------------------------------------
  Dr. Olivier Baujard  /  UIN-CIH Geneve
  Tel : (41 22) 372 62 72
  Fax : (41 22) 372 61 98
  Email : olb@diogenes.hcuge.ch or Olivier.Baujard@imag.fr
  URL: http://expasy.hcuge.ch/sgaico/html/olb/baujard.html

  Interests : DAI, Vision, Classification, Artificial Life
-------------------------------------------------------------

