Assignment: Page 202: Nos. 1,13,14,15,18
a) beta=1/3; p_0=(0,0), p_1=(1,0), p_2=(0,1)
Note: To be consistent with the notation in the book these points can
be thought of as column vectors.
The given data corresponds to the iterated function system
A_0(x,y)=1/3(x.y)
A_1(x,y)=1/3(x-1.y) + (1,0)
A_2(x,y)=1/3(x.y-1) + (0,1)
which gives rise to a type of Sierpinski triangle which, quite
surprisingly, has fractal dimension D=(Log3/Log2). The attractor
of this iterated function system (IFS) is
{(x,y) in KxK | y<= 1-x } where K is the Cantor middle-third set.
b) beta=1/2; p_0=(0,0), p_1=(1,0)
The given data corresponds to the iterated function system
A_0(x,y)=1/2(x.y)
A_1(x,y)=1/2(x-1.y) + (1,0)
Note that the unit interval is fixed with respect to this IFS and hence
, is its attractor with fractal dimension D=(Log2/Log2)=1.
c) beta=1/3; p_0=(0,0), p_1=(1,0), p_2=(0,1), p_3=(1,1)
The given data corresponds to the iterated function system
A_0(x,y)=1/3(x.y)
A_1(x,y)=1/3(x-1.y) + (1,0)
A_2(x,y)=1/3(x.y-1) + (0,1)
A_3(x,y)=1/3(x-1.y-1) + (1,1)
The attractor of this IFS is KxK where K is the Cantor middle-third set,
and its fractal dimension is D=(Log4/Log3), which is exactly twice
that of K.
Yes, C is a fractal with fractal dimension D=(Log4)/Log(15/2).
Yes. Consider the Cantor set that we obtain as a the cartisian product of two Cantor middle-fifths set. Then the fractal dimension would be (by problem 1 and 2)
A set has topological dimension 0 if every point has arbitrarily small neighborhoods whose boundaries do not intersect set. The rationals are such a set since the boundary of a disk with irrational radius fails to intersect the rationals (the sum of a rational and an irrational is irrational).
It is also true that the irrationals have topological dimension 0. To see this, suppose we had the following lemmas:
Lemma Let A be a subset of X. Then,
Lemma For A subset of R^n, t-dimA=n <=> A contains a non-empty open subset of R^n.
Then by the first lemma, the irrationals have t-dim<=1. But by the second lemma the irrationals can not have t-dim=1 since they contain no open subset. Therefore, the t-dim of the irrationals is zero.
Koch's snowflake may be enclosed within a rectangle having length 1 and width 2 sqrt(3)/3, and hence its area is no larger than 2 sqrt(3)/3 square units.
Let A(0) be the area of the equilateral triangle with sides of length 1. Since the height of this triangle is sqrt(3)/2 units, we have that A(0)=sqrt(3)/4 square units. Now at the k-th step of the Koch process, 3*4^(k-1) equilateral triangles each with sqrt(3)/(4*3^(2k)) are added to the snowflake. The combined area of these triangles is
Thus, the area of all these triangles is the sum form k=0 to inf of the previos equality, i.e.,
and so, the area of the Koch snowflake is
square units.
The resulting figure should resemble the Sirpinski triangle as does the figure generated in exercise 17 with slight differences.
Assignment: Page 219: Nos. 4, 7B,C,F, 8A,C, 9, 10, 11.
For the following answer let the symbol @ represent an angle Theta.
First, let z1=r1e^(@1i) and z2=r2e^(@2i). Then
b) F(x+iy)= x^2 + iy^2.
This function is not complex differentiable. Let z(t)= (x_0 +t) +iy_0 and w(t)=x_0+i(y_0+t). Then, (all the limits in this section are taken when the real variable t->0)
lim ((x_0+t)^2+i(y_0)^2-((x_0)^2+i(y_0)^2))/((x_0+t)+iy_0-(x_0+iy_0)) =
lim ((x_0)^2+2tx_0+t^2+i(y_0)^2-(x_0)^2-i(y_0)^2)/(x_0+t+iy_0-x_0-iy_0) =
lim t(2x_0+t)/t = 2x_0
but
lim ((x_0)^2+i(y_0+t)^2-((x_0)^2+i(y_0)^2))/(x_0+i(y_0+t)-(x_0+iy_0)) =
lim ((x_0)^2+i((y_0)^2+2ty_0+t^2)-(x_0)^2-i(y_0)^2)/(x_0+iy_0+it-x_0-iy_0) =
lim it(2y_0+t)/it = 2y_0.
c) F(z) = abs(z).
Since this function is not differentiable in R, then it cannot be
differentiable in C.
f)F(z)=2z(i-z).
Observe that 2z(i-z)=2iz-2z^2, and so F'(z)=21-4z. Indeed, (the limits here are taken as z -> z_0)
lim (2zi-2z_0i-2z^2+2(z_0)^2)/(z-z_0) = lim (2i(z-z_0)-2(z^2-(z_0)^2)/(z-z_0) =
lim ((2i(z-z_0)-2(z+z_0)(z-z_0))/(z-z_0) = lim (2i-2(z+z_0)) = 2i-4z_0.
a) Q_2(z) = z^2 + 2.
The fixed points are
Since Q'_2(z) = 2z,
which is greater than one, and so the fixed point is repelling.
c) F(z)=iz^2.
Since F'(z)=2iz, we have F'(-i)=2i(-i)=-2i^2=2. In other words, this fixed point is repelling. The origin, on the other hand, is superattracting since F'(0)=0.
Since Q_i(i-1) = (i-1)^2+i = i^2+1-2i+i = -i and Q_i(-i)=(-i)^2+i=i-1,these points constitute a 2-cycle for Q_i. Note that Q'_i(z)=2z and
Now, since abs(4(1+i))=4 abs(1+i) = 4 sqrt(2)>1, this 2-cycle is
repelling.
Qo(e^(2PIi/3))=e^(2PIi/3)^2=e^(4PIi/3) and
Qo(e^(4PIi/3))=e^(8PIi/3)=e^(2PI)*e^(2PIi/3)=e^(2PIi/3).
This two cycle is repelling since Qo'(z)=2z and so
(Qo^2)'(e^(2PIi/3)) = Qo'(e^(2PIi/3))*Qo'(e^(4PIi/3)) = 2e^(2PIi/3)*2e^(4PIi/3)
= 4e^(6PIi/3)
and |4e^(6PIi/3)| = 4 > 1.
Qo(e^(2PIi/7))=e^(2PIi/7)^2=e^(4PIi/7) and
Qo(e^(4PIi/7))=e^(8PIi/7) and
Qo(e^(8PIi/7))=e^(16PIi/7)=e^(2PI)*e^(2PIi/7)=e^(2PIi/7)
This two cycle is repelling since Qo'(z)=2z and so
(Qo^3)'(e^(2PIi/7)) = Qo'(e^(2PIi/7))*Qo'(e^(4PIi/7))*Qo'(e^(8PIi/7))
= 2e^(2PIi/7)*2e^(4PIi/7)*2e^(8PIi/7)
= 8e^(14PIi/7)
and |8e^(14PIi/7)| = 8 > 1.