Fractals/Iterations in the complex plane/qpolynomials

Complex quadratic polynomial[1]

Forms

z^2+c

How to compute iteration

In Maxima CAS :

(%i28) z:zx+zy*%i;
(%o28) %i*zy+zx
(%i37) c:cx+cy*%i;
(%o37) %i*cy+cx
(%i38) realpart(z^2+c);
(%o38) -zy^2+zx^2+cx
(%i39) imagpart(z^2+c);
(%o39) 2*zx*zy+cy

Critical point

A critical point of f_c\, is a point  z_{cr} \, in the dynamical plane such that the derivative vanishes :

f_c'(z_{cr}) = 0. \,

Since

f_c'(z) = \frac{d}{dz}f_c(z) = 2z

implies

 z_{cr} = 0\,

One can see that :

  • the only (finite) critical point of f_c \, is the point  z_{cr} = 0\,
  • critical point is the same for all c parameters

z_{cr} is an initial point for Mandelbrot set iteration.[2]

↑Jump back a section

z^2 + m*z

 f_m(z) = z^2 + m z  \,

which has an indifferent fixed point[3] with multiplier[4]

\lambda = m = e^{2 \pi t i} \,

at the origin[5][6]

How to compute iteration

In Maxima CAS :

(%i1) z:zx+zy*%i;
(%o1) %i*zy+zx
(%i2) m:mx+my*%i;
(%o2) %i*my+mx
(%i3) z1:z^2+m*z;
(%o3) (%i*zy+zx)^2+(%i*my+mx)*(%i*zy+zx)
(%i4) realpart(z1);
(%o4) -zy^2-my*zy+zx^2+mx*zx
(%i5) imagpart(z1);
(%o5) 2*zx*zy+mx*zy+my*zx

Critical point

A critical point of f_c\, is a point  z_{cr} \, in the dynamical plane such that the derivative vanishes :

f_m'(z_{cr}) = 0. \,

Since

f_m'(z) = \frac{d}{dz}f_m(z) = 2z + m

implies

 z_{cr} = -\frac{m}{2}\,

One can see that :

  • critical point is related with m value and have to be computed for every m parameters

z_{cr} is an initial point for Mandelbrot set iteration.

Parameter plane lambda

parameter plane

period 1 components

(%i1) e1:z^2+m*z=z;
(%o1) z^2+m*z=z
(%i2) e2:2*z+m=w;
(%o2) 2*z+m=w
(%i3)  s:eliminate ([e1,e2], [z]);
(%o3) [-(m-w)*(w+m-2)]
(%i4) s:solve([s[1]], [m]);
(%o4) [m=2-w,m=w]

It means that there are 2 components of period 1 :

  • one with radius=1 and center=0 ( m=w )
  • second with radius=1 and center= -2 ( m=2-w)


How to compute boundary points of first component :

(%i1) m:exp(2*%pi*%i*p/q);
(%o1) %e^((2*%i*%pi*p)/q)
(%i2) realpart(m);
(%o2) cos((2*%pi*p)/q)
(%i3) imagpart(m);
(%o3) sin((2*%pi*p)/q)

Dynamic plane

1/3

First compute parameter of the function :

/* Maxima CAS session */
(%i1) p:1;
      q:3;
      m:exp(2*%pi*%i*p/q);
(%o1) 1
(%o2) 3
(%o3) (sqrt(3)*%i)/2-1/2
(%i9) float(rectform(m));
(%o9) 0.86602540378444*%i-0.5


Then find fixed points :

/* Maxima CAS session */
(%i10) f:z^2+m*z;
(%o10) z^2+((sqrt(3)*%i)/2-1/2)*z
(%i11) z1:f;
(%o11) z^2+((sqrt(3)*%i)/2-1/2)*z
(%i12) solve(z1=z);
(%o12) [z=-(sqrt(3)*%i-3)/2,z=0]
(%i13) multiplicities;
(%o13) [1,1]

Compute multiplier of the fixed point :

(%i23) d:diff(f,z,1);
(%o23) 2*z+(sqrt(3)*%i)/2-1/2

Check stability of fixed points :

(%i12) s:solve(z1=z);
(%o12) [z=-(sqrt(3)*%i-3)/2,z=0]
(%i20) s:map(rectform,s);
(%o20) [3/2-(sqrt(3)*%i)/2,0]
(%i21) s:map('float,s);
(%o21) [1.5-0.86602540378444*%i,0.0]
(%i24) z:s[1];
(%o24) 1.5-0.86602540378444*%i;
(%i31) abs(float(rectform(ev(d))));
(%o31) 2.645751311064591

It means that fixed point z=1.5-0.86602540378444*%i is repelling.

Second point z=0 is parabolic :

(%i33) z:s[2];
(%o33) 0.0
(%i34) abs(float(rectform(ev(d))));
(%o34) 1.0

Find critical point :

(%i1) solve(2*z+(sqrt(3)*%i)/2-1/2);
(%o1) [z=-(sqrt(3)*%i-1)/4]
(%i2) s:solve(2*z+(sqrt(3)*%i)/2-1/2);
(%o2) [z=-(sqrt(3)*%i-1)/4]
(%i3) s:map(rhs,s);
(%o3) [-(sqrt(3)*%i-1)/4]
(%i4) s:map(rectform,s);
(%o4) [1/4-(sqrt(3)*%i)/4]
(%i5) s:map('float,s);
(%o5) [0.25-0.43301270189222*%i]
(%i6) abs(s[1]);
(%o6) 0.5

1/7

How to speed up computations ? Approximate f^7 by :

f_a^7(z) = (245.4962434402444i-234.5808769813032)*z^8 + z

How to compute f_a^7 :

(%i1) z:x+y*%i;
(%o1) %i*y+x
(%i2) z7:(245.4962434402444*%i-234.5808769813032)*z^8 + z;
(%o2) (245.4962434402444*%i-234.5808769813032)*(%i*y+x)^8+%i*y+x
(%i3) realpart(z7);
(%o3) -234.5808769813032*(y^8-28*x^2*y^6+70*x^4*y^4-28*x^6*y^2+x^8)-245.4962434402444*(-8*x*y^7+56*x^3*y^5-56*x^5*y^3+8*x^7*y)+x
(%i4) imagpart(z7);
(%o4) 245.4962434402444*(y^8-28*x^2*y^6+70*x^4*y^4-28*x^6*y^2+x^8)-234.5808769813032*(-8*x*y^7+56*x^3*y^5-56*x^5*y^3+8*x^7*y)+y

↑Jump back a section

m*z*(1-z)

Description

Critical points

critical points :

  • z = 1/2
  • z = ∞

Parameter plane

period 1 components

(%i1) e1:m*z*(1-z)=z;
(%o1) m*(1-z)*z=z
(%i2) d:diff(m*z*(1-z),z,1);
(%o2) m*(1-z)-m*z
(%i3) e2:d=w;
(%o3) m*(1-z)-m*z=w
(%i4) s:eliminate ([e1,e2], [z]);
(%o4) [m*(m-w)*(w+m-2)]
(%i5) s:solve([s[1]], [m]);
(%o5) [m=2-w,m=w,m=0]

It means that there are 2 period 1 components :

  • discs of radius 1 and centre in 0
  • disc of radius 1 and centre = 2

References

  1. wikipedia : Complex quadratic polynomial
  2. Java program by Dieter Röß showing result of changing initial point of Mandelbrot iterations
  3. wikipedia : fixed point
  4. wikipedia : multiplier
  5. wikipedia : origin
  6. Michael Yampolsky, Saeed Zakeri : Mating Siegel quadratic polynomials.
  7. lambda map at Mu-Ency
↑Jump back a section
Last modified on 5 May 2013, at 13:40