University of Sydney
School of Mathematics and Statistics
Andrew Vincent
Graduate Student, School of Mathematics and Statistics, University of Sydney
Ant Trail Formation and its Application to the Travelling Salesperson Problem
Wednesday, 29th March, 2-3pm, Carslaw 275.
Ants exhibit collective decision making with an apparent absence of
centralised control. One example of this is the laying of pheromone
trails, which allow communication between individuals. When ants find
a couple of food sources, the colony will ``choose'' the ``best''
source to feed at. One interpretation of the ``best'' food source is
the closest.
The Travelling Salesperson Problem (TSP) is to find the shortest path
through a collection of destinations, such that each destination is
visited exactly once.
The connection between ants ``choosing'' the shortest path and solving
the TSP was made about 10 years ago. Since then many algorithms have
been written and tested on various TSPs. Some of these are very good
at finding the shortest path, however as yet no one knows exactly why
they work.
This talk consists of three sections. The first examines a system
which models ants foraging, and concludes with a description of some of
the phenomena of the trail networks of foraging ants. This model is
then reinterpreted to discuss how ants ``choose'' the closest food
source. Finally a reinterpretation is examined which attemps to gain
an insight into how ants solve the TSP.