This page simulates an algorithm for conflict-free coloring of
points with respect to discs, and lets you test the result
yourself.
The problem
Color a set of points in the plane so that every disc containing
at least one point also contains a point whose color is unique in
that disc. The goal is to use as few colors as possible.
It comes from cellular networks: antennas are the points, colors
are frequencies, and a phone anywhere must hear at least one
antenna on a frequency no other nearby antenna uses.
The algorithm
Based on Even, Lotker, Ron and Smorodinsky (2003). While points
remain:
-
Build the Delaunay triangulation of the
remaining points.
-
Color that graph so neighbors differ and keep the largest color
class: an independent set with at least a
sixth of the points.
-
Give that set the next color and remove it.
Any disc holding two or more remaining points contains a Delaunay
edge, so its points never all leave in the same round. That makes
the highest color inside any disc unique. Each round removes a
constant fraction of the points, so only about
log n colors are needed.
Using the simulator
-
Draw points places random points.
- Generate coloring runs the algorithm.
-
Circle: drag a disc and watch its unique color.
-
Rounds replays the algorithm step by step, and
Triangulation shows the graph.