What is it? | |
How does it work? | |
What do we know about it? | |
Where does it come from? | |
What generalizations are possible? |
Other graphs: A way to see the grid is as a set of squares, with some lines ("edges") connecting them, if they are neighbors. We can then extend the idea to other regular grids: take a honeycomb picture, put cells in the junctions of the lines, consider the cells joined by lines as neighbors, and voilà, that's the hexagonal grid. The triangular grid is defined in a similar way (see the picture below). We can define the rule in a general way for arbitrary graphs, if we have a picture of them: the ant enters a node (cell), and turns to the left or to the right, depending on the state it finds there. In a general graph, "turning to the left" means to take the next edge, moving clockwise around the node, and starting with the edge on which the ant was before.
By the way: a simple extension of the system is the inclusion of several
ants; the easiest way to do it is to let the ants run simultaneously on
the grid, ignoring each other (even if they walk over the same sites).
We haven't studied systems with several ants: the dynamics generated by
a single one is complicated enough!
An applet involving multiple ants can be found
here.
What do we know about it?
On the square grid
t = 72 | t = 408 | t = 810 | t = 2600 | t = 15600 |
The third time, the ant arose in the land of physics. There are
several models for the microscopical simulation of fluid dynamics.
In the Lorenz Lattice Gases, a particle moves around between fixed
scatterers, which modify its trajectory and, in some models,
may be modified in turn by the collisions One of this models, the Ruijgrok-Cohen
model, corresponds to the ant. In fact, much of the research on the ant
has been done by Cohen and his co-workers. (Notice that
this is E.G. Cohen, not to be confused with the Jack Cohen we mention
in the Links page.)
As seen above, the first possible generalization is the inclusion
of grids different from the square one. This is a sensible extension,
since there is no special reason for supposing the square lattice to
be the best approximation to two-dimensional interactions; moreover,
as seen above, the dynamics of the system appears to strongly depend on
the topology of the underlying graph. Further extension in this
direction are the inclusion of general bi-regular graphs, or
of general planar graphs (which have been done in our group), or the
consideration of higher dimensions, as has been considered by L. Bunimovich.
On the other hand, the dimension of the grid may also be reduced:
the ant can be defined on the line. Since the resulting dynamics
is rather trivial, several other definitions of the rule have been
considered (some by A. Gajardo in our group, some by L. Bunimovich).
Different definitions of the rule have been explored on the square
lattice too. One way of generalization, which includes several rules,
was considered by Jim Propp, and defines
a rule through a rule-string: instead of going from 0 to 1
and back, a cell chooses its succesive states following a string like,
say, 001011. Some of the resulting ants happen to build highways;
other have increasing bilateral symmetrical patterns like the ones
already found in the hexagonal grid; some seem to be "chaotic"... etc.
Some people have used multiple ants, sometimes with a non-trivial
interaction among them. Further possibilities are the inclusion of some
memory, of more information, etc.
What generalizations are possible?
Of course, the number of possible generalizations is infinite; here
we will only mention the ones that have been already examinated in
some publication.
Last updated: 03-31-2002.
Contact: agajardo@dim.uchile.cl