# CHAOS BASED

COMMUNICATIONS
INTRODUCTION
in
DYNAMICAL SYSTEMS

Alexandru Serbanescu, May, 2008
Dynamical systems I

Variables Linear Nonlinear
n=1

Growth, decay, equilibrium
Exponential growth
RC circuit
Fixed points
Bifurcations
Over damped systems, relaxation dynamics
Logistic equation for single species
n=2

Oscillations
Linear oscillator
Mass and spring
RLC circuit
2-body problem

Pendulum
An harmonic oscillator
Biological oscillators
Predator-prey cycles
Nonlinear electronics (van Der Pol)
n ≥ 3

Civil engineering
Electrical engineering

Chaos
Strange attractors (Lorenz)
3-body problem (Poincare)
Chemical kinetics
Iterated maps (Feigenbaum)
Fractals (Mandelbrot)
Forced nonlinear oscillators (Levison, Smale)

Dynamical systems II

Variables Linear Nonlinear
n>>1

Collective phenomena
Coupled harmonic oscillators
Solid-state physics
Molecular dynamics
Equilibrium statistical mechanics

Coupled nonlinear oscillators
Lasers, nonlinear optics
No equilibrium statistical mechanics
Nonlinear solid-state physics
Heart cell synchronization
Neural networks
Economics

Continuum

Waves and patterns
Elasticity
Wave equations
Electromagnetism (Maxwell)
Quantum mechanics (Schrödinger,
Heisenberg)
Heat and diffusion
Acoustics
Viscous fluids
Spatio-temporal complexity
Nonlinear waves (shocks, solitons)
Plasmas
Earthquakes
General relativity (Einstein)
Quantum field theory
Turbulent fluids
Reaction-diffusion, biological waves
Fibrillation
Epilepsy
Life

Two faces of Chaos

 The word chaos refers to disorder and extreme confusion
 To a scientist, it implies “deterministic disorder”
 This might suggest that a chaotic system should be unpredictable
 On the contrary, a chaotic deterministic system is, in principle, perfectly
predictable
 The sensitive dependence of the system dynamics to the initial conditions
 Chaos is sometimes used as a scapegoat: meteorologists blame chaos for
inaccurate predictions when it is often model inadequacy that is at fault
Deterministic versus Stochastic (1)

 stochastic system

) 1 , 0 ( :
1
N ax x
t t t t
c c + =
+
 deterministic system
) 1 ( 4
1
1
t t t
t t t
z z z
z ay y
÷ =
+ =
+
+
o

Deterministic versus Stochastic (2)

 Deterministic: Everything is described by a single point in state space.
This description completely determines the future.
 Stochastic: Knowledge of present states does not determine
the evolution of future states.
 Can you tell if a system is stochastic or deterministic?
 Should a system be modeled as stochastic or deterministic?
 High dimensional deterministic chaotic system might be
modeled as a stochastic system
 State space trajectory of an autonomous, deterministic system
never crosses itself

Introduction to Dynamical Systems

 A state is an array of numbers that provides sufficient information to describe
the future evolution of the system.
 The collection of these state vectors defines an m-dimensional state space.
 The rule for evolving from one state to another may be expressed as a discrete
map or a continuous flow:
- Map
) (
1 t t
x F x =
+

- Flow
)) ( ( ) ( t x f t x = 

 Fixed point of a map:
) (
0 0
x F x =

 Fixed point of a flow:
0 ) (
0 0
= = x f x

 Non-autonomous system:
) , ( t x f x = 

 Autonomous system:
) (x f x = 

 Dissipative flow
0 < Vf
implies contracting state space volume
 Hamiltonian systems are non-dissipative or conservative,
preserving state space volume (Liouville theorem)
Review of dynamical systems

- State vector
m
x R e in an m-dimensional state space
- Unique description of the state of the system
- Equations of motion:
( )
1
,
n n
x F x a
+
=
provides a rule which determines the evolution of these states
- For a deterministic system this time evolution is unique
- Chaotic systems exhibit sensitive dependence on initial conditions

This implies limited predictability for real systems
Recipe for chaos

- Nonlinearity
- Non-invertible one-dimensional maps
- Sensitive dependence on initial conditions
- Aperiodic behaviour
- Stretching (divergence of neighbouring trajectories)
- Folding (confinement to the bounded space)

Lorenz’s butterfly effect

Demo CHUA generator
DEMO Logistic Map
Periodic solutions
( )
t
u x is a periodic solution of an autonomous system, if for all t:

( ) ( )
t t T +
¬ eI ¬u = u x x x with 0 T >
The classic example of a limit cycle is found in Van der Pol’s
equation:
( )
1 2
2
2 1 2 1
1
x x
x x x x
=
= ÷ ÷

a) limit cycle; b) the time waveform for
1
( ) x t

Some solutions for Duffing’s equation:

1 2
3
2 1 1 2
cos
x x
x x x x t
=
= ÷ ÷ o + ¸ e

Solution of Duffing’s equation for: 0,15; 0,3 şi 1 o = ¸ = e=

Solution of Duffing’s equation for: 0, 22; 0,3 şi 1 o = ¸ = e=
a) A period T=3trajectory; b) the time waveform for
1
( ) x t

Quasi periodic solution
A Van der Pol equation with a limit cycle of period
1
T

( )
( )
1 2
2
2 1 2 1 2
1 cos 2 /
x x
x x x x A t T
=
= ÷ ÷ + t

forced to syncronize with some multiple of the input period
2
T
incomensurate with
1
T (for:
2
0,5; 2 /1,1 A T = = t )

a) The quasi-periodic trajectory; b)The time-waveform of
1
( ) x t
Chaotic trajectory from Duffing’s equation for:

0, 25; 0,3şi 1 o = ¸ = e=

a) the trajectory; b) the time waveform for
1
( ) x t

1/4
Three routes to Chaos

identified for nonlinear dynamical systems:

- The period – doubling sequence
- The intermittency phenomenon
- The breakup of quasiperiodic motion
2/4
The period – doubling route

A sequence of period – doubling bifurcation: period – two to –four
to –eight etc.
The bifurcation values accumulate at the particular parameter
value for which the system becomes chaotic.
k=3 6 1+ = k

3/4
Chaos via intermittency phenomenon

A dynamical system exhibits, over a certain range of parameter
values the following behavior: One or more of its variables,
oscillating periodically in regular and predictable way, suddenly
start to display brief “chaotic burst” of irregular oscillations,
intermittent between long intervals of simple quasiperiodic
motion
¸
|
o
¸
+
¸
-
|
+
|
-
o
+
o
-

4/4
The breakup of quasiperiodic motion

At the certain value of a parameter of a dynamical system you
may notice that the motion turns from periodic to quasiperiodic,
as a second frequency appears which in general is not related with
1
e by a simple rational dependence. In phase space, this would
mean that the orbit now lies on a torus, which intersects a two
dimensional surface of the section along a smooth closed curve.
250 300 350 400 450
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
n
x
n

Bifurcations

- Bifurcation is Latin for “forking into two”, and probably derives from the fact
that in some cases, such as a pitchfork bifurcation, the orbit seems to “fork”
from period one to period two
- A bifurcation is a qualitative change in the dynamics of a system that occurs
when a control parameter is varied
- A bifurcation point is the point in the parameter space where the bifurcation
occurs
- Typically, a bifurcation occurs when an attractor becomes unstable
- Bifurcations are classified according to how stability is lost
- Bifurcation types: saddle node, pitchfork, transcritical, Hopf
- In a map:
o real eigenvalue leaves unit circle at +1: saddle node (or fold)
o real eigenvalue leaves unit circle at -1: pitchfork (or flip)
o conjugate complex eigenvalues leave unit circle simultaneously: Hopf
Bifurcations (continued)

- The particular way in which stability is lost depends on the symmetry
properties of the Jacobian matrix
- Investigate when the eigenvalues of the Jacobian cross the imaginary axis
- Consider the algebraic properties of the Jacobian matrix J:
o An antisymmetric J has pure imaginary eigenvalues
o A symmetric J has pure real eigenvalues
o Complex eigenvalues lead to a rotational flow nearby the fixed point
o Typically perturbations to the fixed point will spiral into the fixed point
before the bifurcation and will spiral out to a limit cycle afterwards
o Real eigenvalues imply an absence of rotation
o Typically, perturbations about such a fixed point flow directly towards the
fixed point before the bifurcation and flow directly outwards to another
equilibrium afterwards

1/6

Consider the logistic equation:

( )
1
1
k k k
x x x
+
= ì ÷ .
This can be considered the discrete part of the continuous logistic
equation:

( ) ( )
1 f x x x = ì ÷ , xe .
The equilibrium point
q
x proves:

( )
q q
f x x = ,
That means:

( )
1
q q q
x x x ì ÷ = .
In this case we have:

1 2
ş
1
0 1 i
q q
x x = = ÷
ì
.
2/6
The equilibrium points
1 2
şi
q q
x x also prove the condition:

( ) ( )
1 2 f x x
'
= ì ÷ ,
So:

( )
( )
1
0
q
f x f
' '
= = ì
and

( )
2
1
1 2
q
f x f
| |
' '
= ÷ = ÷ì
|
ì
\ .
.
The stability condition for
1
0
q
x = is 1 ì < ,
and the stability condition for
2
1
1
q
x = ÷
ì
is 2 1 ÷ì < or: 1 3 < ì < .
3/6

In conclusion;
- if 0 1 sì < , the solution
1
0
q
x = is stable and
2
1
1
q
x = ÷
ì

is unstable;
- if 1 3 sì < , the solution
1
0
q
x = is unstable and
2
1
1
q
x = ÷
ì
is stable;
- if 3 ì > , both
1
q
x and
2
q
x are unstable.
4/6

The second cycle of the dynamic system is characterized by:

( ) ( ) ( ) ( )
( ) ( )
not
2
2
2 3 2
1 1 1
1 1 .
=
f f x f x x x x x
x x x x
= ì ì ÷ ÷ì ÷ = ( ( (
¸ ¸ ¸ ¸ ¸ ¸
= ì ÷ ÷ì ÷

In this case, the equilibrium condition
( )
2
q q
f x x = is:

( ) ( )
2
2 3 2
1 1
q q q q q
x x x x x ì ÷ ÷ì ÷ =
or:

( )
( )
3 4 3 3 2 2 2
2 1 1 0
q q q q
x x x x ÷ì + ì ÷ì +ì ÷ ì ÷ = .
The solutions are:
1
0
q
x = and
2
1
1
q
x = ÷
ì
, and from the
equation: ( ) ( )
2 2
1 1 0
q q
x x ì ÷ì ì + + + ì =

5/6
We have:
( )( )
3,4
1 1 3
2
q
x
ì + ± ì + ì ÷
=
ì
, which are real only
for: 3 ì > .
For the third composition of the iterative logistic function:
( ) ( )
not
3
f f f x f x (
(
¸ ¸
¸ ¸
=
,
we can obtain eight equilibrium points from the equation
( )
3
q q
f x x = , from which four will be stable and four unstable for
3
6 3, 449490... ì = = .
Similarly, for the fourth composition of the iterative logistic
function:

( ) ( )
not
4

q q q
f f f f x f x x
(
(
(
=
¸ ¸
¸ ¸
¸ ¸
=

We will have for
4
3,5440... ì = , 16 fixed points (from which eight
stable and eight unstable), and so on.
6/6

The values
n
ì for which the bifurcations take place ( with 2
n

fixed points, from which
1
2

are stable and
1
2

are unstable )
are:
1 6
2 7
3 8
4 9
5
1 3,568759...
3 3,569296...
3, 449490... 3,569891...
3,545090... 3,569934...
3,564407...
ì = ì =
ì = ì =
ì = ì =
ì = ì =
ì = ...

Feigenbaum shown that these points can be obtained from
the equation:

( ) 1 n
n
c F
÷ ÷
·
ì = ì ÷ · ,
where 3,5699456...
·
ì = , 2,6327... c = , and 4, 6692... F = , which is
Feigenbaum (universal) constant.
Lyapunov exponents

- ( )
1 n n
x F x
+
=
- ( )
0
x ì is Lyapunov exponent
( )
( ) ( )
0
0 0
N x N N
e F x F x
ì
c c = + ÷
( )
( )
0
0
0
1
ln
lim
N
N
dF x
x
N dx
ì
÷·
=
( ) ( )
'
0
1
ln
lim
i
N
x dF x
N
ì
÷·
=
-
( )
0
x
e
ì
average factor separation increases per iteration
- average over attractor
( ) ( )
1
0
' x F x ì µ =
}

- for logistic map
( )
1
0
ln 4 8
ln 2
1
x
dx
x x
ì
t
÷
= =
÷
}