| 
  Home -> Workshops -> Chaos and Fractals Chaos and Fractals: Complex Dynamic SystemsBert G. WachsmuthDepartment of Mathematics and Computer Science
 Seton Hall University
 
 Setup: Suppose  is a family of
  functions, and  . Study the orbits of  : 
 We will study the following two basic question: 
    
      
         
          |  | fix the parameter c, what about the orbits of  for
            various  's |  |  | fix a point  , what about the orbits of  for different values of c |  Example: Take the quadratic family and
  fix c = 0:  . 
    
      Clearly f(0) = 0 and f(1) = 1, so the orbit of 0 is {0, 0, 0 
} and the orbit of 1
      is {1, 1, 1 
}. Also, f(-1) = 1, so the orbit of -1 is {-1, 1, 1, 1 
}.
      Moreover, if  the orbit wil converge to 0, and if  the orbit will converge to positive infinity. Definition: If  is differentiable, and  is a fixed point, i.e.  , then: 
    
      
         
          |  | if  then  is called an
            attracting fixed point |  |  | if  then  is called a
            repelling fixed point |  |  | if  then  is called a
            non-hyperbolic fixed point |  If  is an attracting fixed point then the set of all
      points whose orbit converges to  is called the basin of
      attraction of  Example: Take the quadratic family and
  fix c = -2:  . 
    
      We can easily find the fixed points and periodic points of higher order using Maple
      
 Example: Take  . Then  so that the period points of period n are 
 
    
      
         
          |  | infinitely many period points |  |  | period points are dense in  |  |  | and periodic points are repelling. |  Chaotic Dynamical Systems:  is called
  chaotic if: 
    
      
         
          |  | f has sensitive dependence on initial conditions |  |  | period points are dense in D |  |  | f is topologically transitive (technical) |  Example:  is chaotic (call the function
  f for short):     so that f also has a dense number of repelling periodic points in [-2, 2] Question: How does the quadratic family  go from completely simply (c = 0) to chaotic (c = -2) ? In other words, we now want to
  look at the second question: fix  and vary the parameter.
  The best point to be fixed is the critical point where  
 Bifurcation DiagramExample:  , fix critical point  , vary c from 0 to -2: 
    
      |  | Bifurcation Applet
        |  | Click Controls to set parameters |  |  | Drag mouse to "zoom in" |  |  | Source Code: |  |  Algorithm for Picture: 
    
      
         
          |  | Start with c = 0 |  |  | Compute sequence  without plotting |  |  | Compute sequence  and plot orbits |  |  | Replace c by c - (something) and repeat |  Sarkovskii's Theorem 
    
      Reorder the natural number as follows (Sarkovskii Ordering): 
 
    
      Then if  is continuous, and f has a period point of
      period k, then f also has periodic points of period l for all  in the Sarkovskii ordering. Consequences: 
    
      If f has period 3, then f has periodic points of any period If f has only finitely many periodic points, then these points are powers of 2 If f "wants" to have infinitely many periodic points, it must happen via
      "period-doubling" bifurcation 
 Iterations of Complex FunctionsSame setup as before, but  . Talking about families of
  (analytic) function in C: A family  of analytic functions is normal: 
    
       is equicontinuous
  is locally uniformly bounded
  omits more than 2 points (Montel's Theorem)
 Example:  with c = 0 and z in the complex plane. Inside
  of unit circle and outside of unit circle is where the family is normal. Interior of unit
  circle is the basin of attraction for fixed point z = 0, outside of unit circle is basin
  of attraction for fixed point infinity. In particular, equicontinuity implies that there is no sensitive dependence on initial
  conditions, therefore: normal families are not chaotic. Definition: 
    
       
        |  | Fatou Set:  |  |  | Julia Set:  = complement of Fatou Set |  Originally by Fatou and Gaston Julia, 1920's. Facts about the Fatou Set: 
    
       
        |  | Number of components of F is either infinite or less than or equal to 2 |  |  | F can be decomposed into attracting domains, superattracting domains, parabolic domains,
          Siegel disks, and Herman rings. The dynamics on each of these pieces of F is understood. |  Facts about the Julia Set: 
    
       
        |  |  where  (basin of
          attraction) |  |  | J is the closure of all repelling periodic points |  |  | If the orbits of finite critical points are unbounded, then J is totally disconnected.
          If these orbits are bounded, J is connected |  |  | If  then  |  |  | J is the set where "chaos" takes place |  Algorithm to generate J: 
    
       
        |  | Choose a point  in C |  |  | Iterate  some 50 times, check if orbit is larger than,
          say, 10 |  |  | If orbit is larger than 10, color  black. Otherwise color  according to when it is larger than 10 |  |  | Choose another  and continue |  
 The Mandelbrot SetConsider the function  . Then infinity is an attracting
  fixed point, regardless of c, and 0 is the only critical point. Julia set is the boundary
  of the basin of attraction of infinity, and it is either connected or totally
  disconnected. For this particular family of functions, define: 
 
 Then the boundary of  is the Julia set for this
  particular c, and the set M in parameter space is called the Mandelbrot set. Algorithm to generate M: 
    
      
        Fix  (the critical point). Choose a parameter  Iterate 50 times. If orbit is bounded, color  black. Otherwise, color  depending on when it's bigger than, say, 10. 
    
      |  | Mandelbrot Applet:
        |  | Click Controls to set parameters |  |  | Drag mouse to "zoom in" |  |  | Source Code: |  | It is interesting to put the Mandelbrot set and the Bifucation diagram
  right on top of eachother, using exactly the same scale. There you can see that the big
  cardiot region of the Mandelbrot set corresponds to Julia sets with one attracting fixed
  point, the second circular bulb corresponds to areas with an attracting orbit of period 2,
  and so on. 
 The Henon MapSetup: The family of Henon Maps is defined as  
 Simple Facts: 
    
      
         
          |  | Almost affine map |  |  | Has an inverse map |  |  | WOLOG:  |  |  | If b = 0, then  is topologically conjugate to  |  |  | There is no "distinguished" point (such as the critical point). |  |  | When varying the parameters, we get: |  
           
            |  | If  then H has no fixed points |  |  | If  then H has one real fixed point |  |  | If  then H has two fixed points (a sink and a saddle) |  |  | If  then H has two fixed points (saddles) and a periodic
              two orbit |    Simple Definitions: 
    
      Compute Eigenvalues  of the Jacobian matrix  Note that 
 
    
      If  is a fixed point, then: 
    
      
         
          |  | If  , the fixed point is attracting (and there is a basin
            of attraction), or a sink |  |  | If  , the fixed point is repelling (and there is a basin
            of attraction for the inverse function), or a source |  |  | If  , the fixed point is a saddle point (and there is a
            stable and unstable manifold in the direction of the Eigenvectors corresponding to the
            Eigenvalues) |  |  | If  or  , the fixed point
            is non-hyperbolic (and difficult) |  Theorem: 
    
      There is a combination of (a, b) near b = 0.3 and a = -1.4 where the map is chaotic and
      the iterates has a "strange attractor".  A strange attractor is a set that attracts an open neighborhood, but is not the union
      of sinks. 
    
      |  | Hanon Applet:
        |  | Simple applet to draw iterates of seed point after 100 "pre-iterates" for
          fixed, specified parameters. |  |  | Hit Start to plot |  |  | Change parameters or starting points, hit Start again |  |  | Source Code: |  | 
 Bert G. Wachsmuth (bgw)March 27, 1999
 |