Box Dimension Fractal Dimensions Fractal Dimension Random Fractals. Box Dimension: Set Boxes Linear Size Cover
Box Dimension Fractal Dimensions Fractal Dimension Random Fractals. Box Dimension: Set Boxes Linear Size Cover
Different ways to define fractal dimensions usually lead to the same result.
Important: different ways usually lead to different methods to calculate the fractal
dimension, in particular, in random fractals.
A ε
For a square: N (ε ) = ⇒ df = 2
ε2
For a cube: N (ε ) =
A ε ⇒ df =3
ε3
That is for integer dimensions d f = d as expected !
k
ε = 1/ 3 1
For boxes of size
ε = (1/ 3)2 ε
ε = (1/ 3)3 The number is: N (ε ) = 2k
A ln N (ε ) ln 2k ln 2
Solving 2 = d ⇒ d f =
k
= = ≅ 0.6309
ε 1 k
f
ln ln 3 ln 3
ε
More fractal dimensions
log 4
d
1 1 1 min
d min : l L = l ( L ) = l ( L)
4 5 4
log 5
d min = ≅ 1.161, l = BLdmin
log 4
Shortest path dimension - d min
M l = M (l ) = M (l )
5 7 5
log 7
dl = ≅ 1.209, M (l ) = Cl dl
log 5
= A′l f min
d /d
= Cl dl ⇒ dl = d f / d min
More characteristics of fractals include: backbone, external perimeter, red bonds, etc.
3. Self-affinity
Self-similarity or scale invariance is an isotropic property, the change of scale is the
same in every direction in space.
uuuuuuur
x → 2x
Example: Sierpinski gasket
y → 2y
Self-affinity – include anisotropic symmetry magnifying x in different scale than y.
Example:
n=2 n=3
- Here we see that to get the same picture we need to magnify the x axis by
4 and y axis by 2, x → 4 x, y → 2 y
1 1 1
M Lx , L y = M ( Lx , Ly )
4 2 4
Generalization of self-similar fractals: M (bL) = b M (L)
f d
3.1 Fractal dimension – self-affine structures
d xf
1 1 1 1
M Lx , Ly = M ( Lx , L y ) = M ( Lx , Ly )
4 2 4 4
d fy
1 1 1 1
M Lx , Ly = M ( Lx , L y ) = M ( Lx , Ly )
4 2 4 2
d xf
1 1
= ⇒ d xf = 1,
4 4
d fy
1 1
= ⇒ d fy = 2
2 4
Example: self-affine Sierpinski carpet
d xf
1 1 1 1
M Lx , Ly = M ( Lx , Ly ) = M ( Lx , Ly )
3 2 3 3
d fy
1
= M ( Lx , Ly )
2
log 3
d xf = 1, d fy =
log 2
Self affine fractals
Here also
d xf
1 1 1 1
M Lx , Ly = M ( Lx , Ly ) = M ( Lx , Ly )
3 2 3 3
dy
1 f log 3
= M ( Lx , Ly ), d fx = 1, d fy =
2 log 2
Generalization:
Start with a square of unit size:
Divide x axis to b1 and y axis to b2
(a)
1 1 1
For size we need 3⋅ ⋅
3 3 2 = rectangular area × number of rectangulars
2
1 box area
3
1 1
3k ⋅ ⋅ k
1 k
3 2 boxes. More general: if we divide to
In general for k we need N (ε ) = 2
3 1
k
3
b1 × b2 rectangulars and leave n of them full, we obtain a box of size ε = b1− k and the
−k −k
k b1 ⋅ b2
number of boxes N (ε ) = n
( )
b1−k 2
nb1 nb
ln 1
k ln
−dl ln N (ε ) b2 b1
The local box dimension: N (ε ) = ε f , d lf = = =
1 k ln b1 ln b1
ln
ε
Self affine curves – single valued
Example:
α
Dimension α defined by W (bL) = b W ( L) L
- The fractal dimension of the random Sierpinski carpet is the same as the
d
1 1 1 f log8
deterministic: M L = M ( L) = M ( L), d f = ≅ 1.893
3 8
3 log 3
- The self-similarity is not exact – valid statistically
Random Fractals – Fractal Dimension
−d f
N (ε ) = Aε
log N (ε ) = log A − d f log ε
−α
For isotropic fractals we expect C (r ) = C (r ) = Ar .
The mass within a radius R is:
R
M ( R) = ∫ C (r )d d r = R −α + d ≡ R
df
⇒ α =d −df
Since physical fractals have lower and upper bounds length scales
( λ− and λ+ )
4π 4π
It follows that only for sin α < q < sin ϑ , we obtain d f
λ+ λ−
- Measurements of S (q ) yields d f
Example: polymers.