The Central Idea
A fractal is a geometric or mathematical object whose structure is generated or revealed through repeated scaling, recursion, or iteration.
Many fractals exhibit self-similarity: smaller parts resemble the whole.
The central pattern is:
\[ \boxed{\text{simple rule}+\text{repetition across scales}=\text{complex form}}. \]Self-Similarity
A figure is self-similar when parts of the figure resemble the entire figure after scaling.
The resemblance may be:
- exact;
- approximate;
- statistical.
Exact Self-Similarity
In exact self-similarity, smaller copies reproduce the whole according to an exact mathematical rule.
Classical examples include:
- the Cantor set;
- the Koch curve;
- the Sierpinski triangle.
Approximate Self-Similarity
A natural or constructed form may resemble itself across scales without matching perfectly.
Branching systems, coastlines, clouds, and rough surfaces can display approximate self-similar behavior over limited scale ranges.
Statistical Self-Similarity
In statistical self-similarity, exact copies are not required.
Instead, statistical properties remain similar under changes of scale.
This viewpoint is useful for irregular natural patterns and stochastic models.
Iteration
Iteration means repeatedly applying a rule.
If
\[ x_{n+1}=F(x_n), \]then the same transformation \(F\) is applied again and again.
Fractal geometry often translates such repeated rules into spatial form.
Recursive Construction
A recursive geometric construction defines each stage from the previous stage.
We may write
\[ S_{n+1}=T(S_n), \]where \(T\) is the construction rule.
The limiting object, when a suitable limit exists, is the fractal.
Finite Stage Versus Limiting Fractal
A computer drawing can display only finitely many iterations.
The mathematical fractal is often defined as the limiting object after infinitely many stages.
Therefore:
\[ \boxed{\text{finite approximation}\neq\text{ideal limiting fractal}}. \]The Cantor Set
Begin with the interval
\[ [0,1]. \]Remove the open middle third:
\[ \left(\frac13,\frac23\right). \]Then remove the middle third of each remaining interval, and continue indefinitely.
First Stages of the Cantor Construction
Stage \(0\):
\[ [0,1]. \]Stage \(1\):
\[ \left[0,\frac13\right] \cup \left[\frac23,1\right]. \]Stage \(2\) contains four intervals, each of length
\[ \frac19. \]Number and Scale of Cantor Pieces
At each stage:
- each surviving interval produces \(2\) copies;
- each copy is scaled by a factor of \(\frac13\).
After \(n\) stages, there are
\[ \boxed{2^n} \]intervals, each of length
\[ \boxed{3^{-n}}. \]Total Length at Finite Cantor Stages
At stage \(n\), total interval length is
\[ 2^n\cdot3^{-n}. \]Therefore,
\[ \boxed{ L_n= \left(\frac23\right)^n }. \]As
\[ n\to\infty, \]this total length approaches zero.
The Cantor Set Is Not Empty
Even though the remaining total interval length tends to zero, infinitely many points survive.
In fact, the Cantor set is uncountable.
This is an important example of how size can depend strongly on what notion of size we use.
The Koch Curve
Begin with a line segment.
Divide it into three equal parts.
Replace the middle third by two sides of an outward equilateral triangle.
Each segment is therefore replaced by four smaller segments.
Koch Scaling Rule
Each segment produces
\[ N=4 \]smaller copies.
Each copy has scale factor
\[ r=\frac13. \]This simple rule determines both its length growth and self-similar dimension.
Length Growth of the Koch Curve
Each stage multiplies total length by
\[ \frac43. \]If the original length is \(L_0\), then
\[ \boxed{ L_n= L_0\left(\frac43\right)^n }. \]Therefore,
\[ L_n\to\infty. \]The Koch Snowflake
Apply the Koch construction to each side of an equilateral triangle.
The result is the Koch snowflake.
Its boundary length grows without bound while the enclosed area approaches a finite limit.
Infinite Boundary, Finite Area
The Koch snowflake demonstrates that:
\[ \boxed{\text{infinite perimeter does not imply infinite area}}. \]This challenges intuition built from ordinary smooth polygons and curves.
The Sierpinski Triangle
Begin with a filled equilateral triangle.
Divide it into four congruent smaller triangles and remove the central one.
Repeat the process on each remaining triangle.
Sierpinski Scaling
Each stage produces
\[ N=3 \]self-similar copies, each scaled by
\[ r=\frac12. \]This immediately suggests a dimension between \(1\) and \(2\).
Area of Sierpinski Stages
Each stage retains
\[ \frac34 \]of the previous area.
Therefore,
\[ \boxed{ A_n= A_0\left(\frac34\right)^n }. \]Hence
\[ A_n\to0. \]Self-Similar Dimension
Suppose a self-similar object consists of \(N\) copies of itself, each scaled by a factor \(r\), where
\[ 0Solving for Self-Similar Dimension
From
\[ Nr^D=1, \]we obtain
\[ r^D=\frac1N. \]Taking logarithms:
\[ D\ln r=-\ln N. \]Therefore,
\[ \boxed{ D= \frac{\ln N}{\ln(1/r)} }. \]Dimension of the Cantor Set
For the Cantor set,
\[ N=2, \qquad r=\frac13. \]Thus
\[ \boxed{ D= \frac{\ln2}{\ln3} }. \]Numerically,
\[ D\approx0.631. \]Dimension of the Koch Curve
For the Koch curve,
\[ N=4, \qquad r=\frac13. \]Therefore,
\[ \boxed{ D= \frac{\ln4}{\ln3} }. \]This is approximately
\[ 1.262. \]Dimension of the Sierpinski Triangle
Here
\[ N=3, \qquad r=\frac12. \]Thus
\[ \boxed{ D= \frac{\ln3}{\ln2} }. \]Numerically,
\[ D\approx1.585. \]Why Fractal Dimension Can Be Noninteger
Ordinary geometric objects have familiar dimensions:
- point: \(0\);
- curve: \(1\);
- surface: \(2\);
- solid region: \(3\).
A fractal can scale in a way that lies between these ordinary categories.
Its scaling dimension can therefore be noninteger.
Dimension Measures Scaling, Not Visual Thickness Alone
Fractal dimension should not be interpreted merely as visual appearance.
It captures how the amount of structure changes as measurement scale changes.
Box-Counting Dimension
A broader notion of fractal dimension uses boxes of side length
\[ \varepsilon. \]Let
\[ N(\varepsilon) \]be the number of boxes needed to cover the set.
The box-counting dimension, when the limit exists, is
\[ \boxed{ D= \lim_{\varepsilon\to0} \frac{\ln N(\varepsilon)} {\ln(1/\varepsilon)} }. \]Why Box Counting Is Useful
Exact self-similarity is convenient but restrictive.
Box counting can be applied to more general sets, including irregular geometric data and numerical approximations.
Measurement Depends on Scale
Fractals make scale dependence especially visible.
When a rough curve is measured with smaller and smaller rulers, more detail may be detected.
The measured length can therefore increase as the measurement scale decreases.
The Coastline Idea
A coastline measured with a long ruler ignores small bays and inlets.
A shorter ruler follows more of the fine detail.
This motivates scale-dependent models of irregular boundaries.
Real coastlines are not exact mathematical fractals at all scales, but the fractal viewpoint can describe their roughness over useful ranges.
Iterated Function Systems
An iterated function system uses a finite collection of transformations
\[ T_1,T_2,\ldots,T_N. \]These transformations repeatedly map a set into smaller copies.
Many classical self-similar fractals can be described this way.
Contractions
A contraction shrinks distances.
Informally, if
\[ d(T(P),T(Q)) \le r\,d(P,Q) \]with
\[ 0Affine Fractal Construction
Many iterated-function-system transformations are affine:
\[ \boxed{ T(\mathbf x) = A\mathbf x+\mathbf b }. \]The matrix \(A\) can scale, rotate, reflect, or shear, while \(\mathbf b\) translates.
This connects fractals directly with transformation geometry.
Recursive Trees
A simple branching fractal begins with one segment.
At the end of each segment, shorter rotated branches are added.
Repeating this rule produces a recursive tree.
The geometry is controlled by:
- branch angle;
- length scale factor;
- number of branches;
- iteration depth.
The Mandelbrot Set
The Mandelbrot set is defined through iteration in the complex plane.
Start with
\[ z_0=0 \]and iterate
\[ \boxed{ z_{n+1}=z_n^2+c }. \]The complex number \(c\) belongs to the Mandelbrot set when the resulting sequence remains bounded.
Complex Coordinates
Write
\[ c=a+bi. \]Then \(c\) corresponds to the point
\[ (a,b) \]in the complex plane.
The Mandelbrot set is therefore a planar set of parameter values.
Julia Sets
For a fixed complex parameter \(c\), consider the same iteration
\[ z_{n+1}=z_n^2+c \]but vary the starting value \(z_0\).
The associated Julia set describes the boundary between different long-term behaviors of those starting points.
Mandelbrot and Julia Sets Are Related but Different
The Mandelbrot set organizes parameter values \(c\).
A Julia set is associated with one chosen parameter \(c\) and organizes starting values \(z_0\).
They arise from the same iteration but answer different questions.
Escape-Time Computation
Computer images of the Mandelbrot set commonly use an escape test.
Iterate
\[ z_{n+1}=z_n^2+c \]and monitor \(|z_n|\).
For this quadratic iteration, once
\[ |z_n|>2, \]the sequence will escape to infinity.
Finite Computation Versus Exact Membership
If a point escapes during computation, we can conclude it is outside the Mandelbrot set.
If it has not escaped after finitely many iterations, that alone does not prove exact membership.
Numerical rendering therefore approximates the ideal mathematical set.
Fractal Boundaries
Fractal behavior often appears most strongly on boundaries.
A region may have an ordinary two-dimensional interior while its boundary exhibits intricate scaling structure.
Fractals in Three Dimensions
Fractal constructions can extend into three-dimensional space.
Examples include:
- the Menger sponge;
- recursive tetrahedral constructions;
- fractal terrain;
- three-dimensional iterated-function systems.
The Menger Sponge
Begin with a cube divided into
\[ 3\times3\times3=27 \]smaller cubes.
Remove selected central cubes and repeat the same pattern on every remaining cube.
The standard Menger sponge retains \(20\) of the \(27\) subcubes at each stage.
Dimension of the Menger Sponge
Each retained copy is scaled by
\[ r=\frac13, \]and there are
\[ N=20 \]copies.
Therefore,
\[ \boxed{ D= \frac{\ln20}{\ln3} }. \]This dimension lies between \(2\) and \(3\).
Fractal Dimension and Spatial Filling
A dimension between \(1\) and \(2\) suggests a set more complicated than a smooth curve but not area-filling in the ordinary sense.
A dimension between \(2\) and \(3\) suggests a set more intricate than a smooth surface but not volume-filling in the ordinary sense.
Self-Similarity and Symmetry
Ordinary symmetry compares a figure with transformed copies at the same scale.
Self-similarity adds scaling:
\[ \boxed{\text{shape similarity across different scales}}. \]This makes scaling itself part of the structural symmetry.
Scale Invariance
A mathematically ideal self-similar fractal may reproduce its structural rule indefinitely under magnification.
This is a form of scale invariance.
Physical objects, however, usually have lower and upper scale limits.
Fractals and Power Laws
Scaling behavior often takes a power-law form.
For example,
\[ N(\varepsilon) \approx C\varepsilon^{-D}. \]Taking logarithms gives
\[ \ln N \approx \ln C + D\ln(1/\varepsilon). \]Thus dimension can appear as a slope on a log-log plot.
Fractals and Discrete Approximation
Fractal constructions connect continuous-looking form with discrete iteration.
Each stage is finite and combinatorial.
The limiting object emerges from the behavior of the entire infinite sequence.
Fractals and Meshes
Fractal-like surfaces can be approximated using meshes.
Recursive subdivision can add smaller geometric features at each stage.
This connects fractal generation with the mesh and tessellation methods studied in neighboring lessons.
Fractals in Modeling
Fractal ideas can be useful for modeling:
- terrain;
- branching structures;
- porous materials;
- rough boundaries;
- cloud-like textures;
- multiscale networks.
Fractal Models Are Approximations
Calling a natural object fractal does not usually mean it possesses exact mathematical self-similarity through infinitely many scales.
A fractal model captures selected scaling behavior over a relevant range.
Common Error: Every Irregular Shape Is a Fractal
Irregularity alone is not enough.
Fractal classification involves structural scaling, recursion, dimension, or related multiscale behavior.
Common Error: Self-Similar Means Identical at Every Scale
Exact mathematical fractals may have exact self-similarity.
Other fractal models exhibit approximate or statistical self-similarity.
These distinctions matter.
Common Error: Fractal Dimension Must Be an Integer
Fractal dimensions are often noninteger.
For the Koch curve,
\[ D=\frac{\ln4}{\ln3}\approx1.262. \]Common Error: Infinite Perimeter Means Infinite Area
The Koch snowflake has infinite limiting perimeter but finite limiting area.
Length and area measure different geometric quantities.
Common Error: Zero Length Means No Points
The Cantor set has zero total interval length in the usual one-dimensional measure, yet it contains uncountably many points.
Measure and cardinality are different concepts.
Common Error: A Finite Rendering Is the Exact Fractal
A displayed or computed image contains finite resolution and finite iteration.
The mathematical fractal is often defined through an infinite limiting process.
Common Error: Dimension Formula Works for Every Self-Similar Picture
The simple equation
\[ Nr^D=1 \]applies cleanly to suitable self-similar constructions.
Overlapping copies, unequal scale factors, or more complicated structures can require more advanced analysis.
A Reliable Procedure for an Exact Self-Similar Fractal
- Identify the initial geometric object.
- State the replacement or iteration rule.
- Count the number \(N\) of self-similar pieces.
- Determine the scale factor \(r\) of each piece.
- Track length, area, volume, or count across stages.
- Write a recurrence or explicit sequence.
- Find limiting behavior when appropriate.
- Compute self-similar dimension using \(Nr^D=1\) when valid.
- Distinguish the finite stages from the limiting set.
- Check whether the construction is exact, approximate, or statistical.
Example: Full Koch Analysis
Each segment is replaced by four segments of one-third the original length.
Thus
\[ N=4, \qquad r=\frac13. \]Total length after \(n\) stages is
\[ L_n=L_0\left(\frac43\right)^n. \]The limiting length diverges:
\[ L_n\to\infty. \]The self-similar dimension is
\[ \boxed{ D=\frac{\ln4}{\ln3} }. \]Example: Full Sierpinski Analysis
Each stage retains three triangles, each scaled by one-half.
Thus
\[ N=3, \qquad r=\frac12. \]Area after \(n\) stages is
\[ A_n=A_0\left(\frac34\right)^n, \]so
\[ A_n\to0. \]Its self-similar dimension is
\[ \boxed{ D=\frac{\ln3}{\ln2} }. \]Practice 1: Cantor Pieces
How many intervals remain after \(5\) stages of the Cantor construction?
Practice 2: Cantor Interval Length
What is the length of each surviving interval after \(4\) stages?
Practice 3: Koch Segment Count
Starting from one segment, how many segments are present after \(3\) Koch iterations?
Practice 4: Koch Length
If the original segment has length \(9\), what is the total length after \(2\) Koch iterations?
Practice 5: Sierpinski Triangle Count
How many retained small triangles appear after \(4\) iterations?
Practice 6: Cantor Dimension
Write the exact self-similar dimension of the Cantor set.
Practice 7: Sierpinski Dimension
Write the exact self-similar dimension of the Sierpinski triangle.
Practice 8: Menger Dimension
A self-similar construction has \(20\) copies, each scaled by \(\frac13\). Write its self-similar dimension.
Practice 9: Mandelbrot Iteration
State the standard quadratic iteration used to define the Mandelbrot set.
Practice 10: Finite Versus Limiting Object
Is a stage-\(8\) drawing of the Koch curve the exact mathematical limiting fractal?
Practice Answers
Practice 1
The number of intervals is
\[ 2^5=\boxed{32}. \]Practice 2
Each stage multiplies interval length by \(\frac13\):
\[ \boxed{3^{-4}=\frac1{81}}. \]Practice 3
Each segment produces four new segments:
\[ 4^3=\boxed{64}. \]Practice 4
Use
\[ L_n=L_0\left(\frac43\right)^n. \]Thus
\[ L_2 = 9\left(\frac43\right)^2 = 9\cdot\frac{16}{9} = \boxed{16}. \]Practice 5
Each retained triangle produces three:
\[ 3^4=\boxed{81}. \]Practice 6
\[ \boxed{ D=\frac{\ln2}{\ln3} }. \]Practice 7
\[ \boxed{ D=\frac{\ln3}{\ln2} }. \]Practice 8
\[ \boxed{ D=\frac{\ln20}{\ln3} }. \]Practice 9
\[ \boxed{ z_{n+1}=z_n^2+c, \qquad z_0=0 }. \]Practice 10
No.
It is a finite approximation to the limiting Koch curve.
What to Remember
- Fractals arise through iteration, recursion, scaling, or multiscale structure.
- Self-similarity can be exact, approximate, or statistical.
- For suitable exact self-similar constructions, \[ Nr^D=1 \] and \[ D=\frac{\ln N}{\ln(1/r)}. \]
- The Cantor set has dimension \[ \frac{\ln2}{\ln3}. \]
- The Koch curve has dimension \[ \frac{\ln4}{\ln3}. \]
- The Sierpinski triangle has dimension \[ \frac{\ln3}{\ln2}. \]
- Finite computer images approximate ideal fractals defined by infinite processes.
- Fractal ideas connect geometry with sequences, transformations, dynamical systems, probability, and computational modeling.
The central habit is: identify the repeating rule, the scale factor, and what geometric quantity changes from one iteration to the next.
Where We Go Next
Fractal geometry shows how complex spatial form can emerge from repeated subdivision and scaling. The next lesson turns to discrete surface representations built from connected polygonal pieces.
In Meshes and Tessellated Surfaces, we will study triangulated and polygonal meshes, adjacency, refinement, approximation, topology, and the representation of curved surfaces through discrete elements.