next up previous
Next: Fractal Approximation of Sets Up: No Title Previous: Introduction

   
Iterated Function Systems

Following M. Barnsley [2], under fractals we mean compact subsets of a complete metric space.

A deterministic Iterated Function System (IFS) [2] is a structure

F = ((X,d), f1, f2, ..., fN), (1)

where (X,d) is a complete metric space, with metric d, $f_i:X \rightarrow X$ - a contiguous function $\forall i$. An IFS F induces an operator $F:{\cal H}(X) \rightarrow {\cal
H}(X)$, where ${\cal H}(X)$ - the space of all compact subsets of X:

\begin{displaymath}F = \bigcup_{i=1}^q f_i.
\end{displaymath} (2)

The metric space $({\cal H}(X),h)$, with Hausdorff metric h, is also complete. When functions fi are contractions, the IFS is hyperbolic. In this case, there is a compact subset AF of X which is the fixed point of the operator F:


\begin{displaymath}\exists A_F \in {\cal H}(X): \; A_F = F(A_F).
\end{displaymath} (3)

Such a set is called the attractor of the IFS F. The pair $({\cal H}(X), F)$ is a set dynamical system whose attractive point is a set (rather than a point) - the attractor AF. Realizing the dynamics of such a dynamical system, we can build a fractal set.

The Hausdorff metric is defined as follows:


\begin{displaymath}h(A,B)=\max(d_{\rm s}(A,B),d_s(B,A)),
\qquad A,B \in {\cal H}(X),
\end{displaymath} (4)

where


\begin{displaymath}d_{\rm s}(A,B)=\max_{x \in A} \min_{y \in B} (d(x,y)).\end{displaymath}

We are dealing here with hyperbolic IFSs with all fi affine transformations.


next up previous
Next: Fractal Approximation of Sets Up: No Title Previous: Introduction
IMACS ACA'98 Electronic Proceedings