7.3. Fractal Geometry and Quotient Space Analysis

7.3.1. Introduction

A famous Sierpinski carpet is shown in Fig. 7.2. It’s a typical fractal graph. From the quotient space view point, it’s a chain of quotient spaces, where image has one element, image has four and image has 13 elements. Therefore, the concepts of quotient space and fractal geometry have a close relation. We will discuss them using the quotient space approximation principle in the following sections.

7.3.2. Iterated Function Systems

In order to investigate fractal graphs from quotient space theory, the concept of quotient fractals is established first. Then, the quotient fractals are used to approximate fractal graphs. Its procedure is the following. An equivalence relation is defined through a mapping. A corresponding chain of hierarchical quotient spaces is built by the equivalence relation. Then, quotient mappings are set up on the quotient spaces. Finally, quotient fractals are obtained from the quotient mappings.
image
Figure 7.2 A Sierpinski Carpet

Definition 7.19

A set image of compressed mappings on a compact metric space image is given, where compression factor image. Let image. It is called an iterated function system on image, simply denoted by image.

Definition 7.20

Assume that image is a complete metric space. image, image, where H(x) is a power set of image. Define image. image is called a distance from point image to set image.

Definition 7.21

Assume that image is a complete metric space. Sets image, define image. image is called a distance from set image to set image.

Definition 7.22

image is a complete metric space. The Hausdorff distance between points A and B on H(x) is defined as image.
The main theorem in iterated function systems is the following.

Theorem 7.4 (Attractor Theorem)

Assume that image is an iterated function system on X. W is a fractal mapping on X. Then, image holds on (H(x),h(d)). Namely, there exists a unique fixed point P on H(x), i.e.,

image

P is a corresponding fractal graph on iterated function system IFS=W.

7.3.3. Quotient Fractals

A mapping is used to define an equivalence relation as follows.

Definition 7.23

An iterated function system image on image is given. Construct image and image. Letting image, then image is a quotient space of image.
Now, image is partitioned as follows, i.e., image is partitioned as image. Let

image

We obtain

image(7.16)

Set image remains unchanged.
Then, we have a partition of image denoted by image. From Formula (7.16), it’s known that image is a quotient space of image.
By induction, assume that image is a known quotient space. For its elements, the partition procedure is the following. Let

image

We obtain

image

Obviously, image image compose a chain of hierarchical quotient spaces. For simplicity, the element image of image is denoted by its subscript image in the following discussion.

Definition 7.24

Define a mapping image on image as image, where image is a nature projection and image.

Definition 7.25

Define image on image, where image is an invariant subset on image corresponding to mapping W, i.e., a fractal graph.

Definition 7.26

image is called a quotient fractal model of iterated function system image.

Theorem 7.4

If an iterated function system IFS on image is given, then it corresponds to a chain of hierarchical quotient sets on image.

Proof

Assume that mapping image. Let image and image.
(1) Assume that image. Then, image compose a partition of X. Its corresponding quotient space is denoted by image. Again, let image and image. Then, image compose a partition of image denoted by image. Obviously, image is a quotient space of image.
By induction, define

image

Let

image

Obviously, image compose a partition of X denoted by image. It’s easy to show that image is a quotient space of image.
Therefore, image is a chain of quotient sets of X or a chain of quotient spaces corresponding to an iterated function system.
(2) When some image and image are overlapping, i.e., image. An abstract space image can be constructed, where element i corresponds to set image. Define an abstract space image, where element image corresponds to set image, i.e., image.
Generally, define image, where element image corresponds to image, image. Let image. The point on image is an infinite sequence composed by 0,1,…,n, i.e., image.
Similarly, we may have a chain image of quotient sets.
We have a profound relation between quotient fractals and fractal graphs as follows.

Theorem 7.5 (Quotient Fractal Approximation Theorem)

We have the following properties.

Property 1

image

Property 2

image is an invariant subset on image corresponding to mapping image.

Property 3

image is an iterated function system on image. Then, image, its convergence is based on the Hausdorff distance defined on the power set of image.

Property 4

image

7.3.4. Conclusions

Property 3 in Section 7.3.3 is the quotient fractal approximation theorem of fractal graphs. It means that in fractal geometry we can still use a set of simple quotient spaces to approximate the original space so that the computational complexity is reduced. This is just the basic principle of quotient space approximation method and the multi-granular computing as well.
..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
3.128.173.53