Nonlinear Dynamic

1
Dynamic systems
Differential equations:
Describe the evolution of the system in
continuous time.

(1)


Iterated maps:
Describe the system in discrete time
2
Linear and nonlinear systems
The system (1) is linear, if all xi (x1.xn)
appear to the first power only. Otherwise the
system is nonlinear (products, powers, &
Cos(x)..)

Linear Easy to solve analytically, (system
can be broken into parts).

Nonlinearity difficult to solve analytically.
3
Flows on the line
4
1D order system,
For this system:
: is the position of an imaginary particle
moving along the real line.
: is the velocity of the particle.
System vector field on the line, and to draw
this vector we need to plot vs (Phase
portrait).

x

x
x
x
5
The phase portrait
The phase portrait of a dynamical system
contains a lot of information about the
solutions of the system, without all the detail
one finds in a solution formula.
- Draw the phase space.
- Find the equilibrium points.
- Draw them in the phase space.
- Determine the motion in between any two
equilibrium points.

0 = x

6
Motion between Fixed Points F.P
Imagine that a fluid is flowing steadily along x-
axis with velocity equal to f(x) then:
- The flow is to right when > 0.
- The flow is to left when < 0.
- There is no flow when = 0. (F.Ps)
Then the system looking to flow (along
trajectory) to a stable position from right or
left.
x

7
y y sin =

Draw the phase space:

Find the equilibrium points:
0 sin = y
,..... 2 , , 0 t t = y
Draw them in the phase space
Motion in between F.P
8
Trajectory and phase space
IC
(Position, Velocity)
Trajectory
Phase space
Our goal is to solve the system by drawing the trajectories without actually
solving the system
9
Stability Definitions
A fixed point is called stable if solutions that
start near it stay near it.
(flow from left & right toward F.P)

A fixed point is called unstable if solutions that
start near it, end up wandering away from it.
(flow from left & right away from F.P)

A fixed point is called asymptotically stable if
solutions that start near it, approach the fixed
point as t ---> .

10
Flow , 0 > x

Flow ,
0 < x

11
Stability of equilibrium solutions
F.Ps
Stable
Attractors or Sinks
Flow toward them
Solid black dots
Unstable
Repellers or Sources
Flow outward them
Open circles
12
- Find all fixed Points (F.Ps), by letting f(x)=0.

- Plot f(x), and show all F.Ps

- Specify the flow direction.

- Classify the stability for each F.P
Procedures to analyze stability of F.Ps
13
Linear stability analysis
Its a way to classify the stability of the F.Ps
by introducing a small perturbation (t) away
from F.P and check if it grows or decays:
- (t) grow if Unstable
- (t) decay if Stable
Or in other word:
- (+ve) slope Unstable.
- (-ve) slope Stable.



*) (x
14
Linear stability analysis limitation
The linearization method cannot be used :

- If f(x) has a vertical tangent at F.P.
- If f(x) has a horizontal tangent at F.P.

In both cases f(x) cannot be expanded in a
Taylor series near x = x* since there is no
Taylor series.

15
Potentials
In term of potential energy 1D system is given
by:

Potentials always decreases along
trajectories Particle always moves toward
lower potentials.
Local minima of V(x) Stable F.P
Local maxima of V(x) Unstable F.P

dx
dV
=
16
Notes on 1D line systems
If f(x) is smooth enough, then the solution
exist and unique.
The approach to equilibrium always
monotonic.
If we flow monotonically on a line, we will
never come back to starting point, so
oscillations can never occur. (there are no
periodic solutions)

17
Bifurcations in one-dimensional
systems
Bifurcation theory is the study of how solutions
of dynamical systems change as a parameter is
changed.

For a dependence parameter T:
If the phase portraits for T < Tc and T > Tc are
qualitatively different, we say that a bifurcation
occurs at T = Tc.

The qualitative change lead to create or destroy
of F.Ps or change in there stability.
( ) x f x =

18
Bifurcation
Bifurcation
Static
Saddle-Node Transcritical Pitchfork
Dynamic
Hopf
19
The saddle-node bifurcation
As the r varied stable and unstable F.Ps meet each
other and both of them destroyed or vanish.

There are three different possibilities:
(a) , there are two distinct fixed points:
(b) , there's only one fixed point:
(c) , theres no fixed points (F.P cannot be imaginary)

2
x x =o

o o = = +
2 2
0 x x
0 < o
o = * x
0 = o
0 *= x
0 > o
20
The saddle-node bifurcation
21
The Transcritical bifurcation
Two solutions stable and unstable when they
met they exchange their stability stable one
become unstable and the unstable one
become stable.
If < 0 :
- x* = 0 ----> Stable.
- x *= ----> Unstable.
If > 0:
- x* = 0 ----->Unstable
- x *= ----> Stable



2
x x x =o

o o = = = * , 0 * 0
2
x x x x
22
The supercritical pitchfork bifurcation
Common in physical problems that have symmetry.
F.P tend to appear and disappear in symmetrical
pairs.
There are three different possibilities:
(a) , (F.P cannot be imaginary) , so there's
only one fixed point: Stable
(b) , one fixed point: Stable
(c) , there are three fixed points:




3
x x x =o

o = = * , 0 * x x
0 < o
0 = o
0 > o
0 *= x
0 *= x
= 0 * x Unstable
= o * x Stable
23
The subcritical pitchfork bifurcation
This kind of bifurcation is dangerous.
There are three different possibilities:
(a) , (F.P cannot be imaginary), only one
fixed point: Unstable
(b) , only one fixed point: Unstable
(c) , there are three fixed points:




3
x x x + =o

o = = * , 0 * x x
0 > o
0 = o
0 < o
0 *= x
0 *= x
= 0 * x
Unstable = o * x
Stable
24
Flows on the circle
The main difference with flows on the line is that
now the flow can return to where it was, by going
around the circle. Thus, periodic solutions are
possible!

flow CCW.

flow CW.

0 > u

0 < u

) (u u f =

25
The uniform oscillator
The simplest flow on the circle is given by the
system:

This system is easily solved:

All points are back where
they started after period of:

e u =

0
u e u + = t
e
t 2
= T
26
The non-uniform oscillator
More complicated system:
Fixed points:
There are three different cases:
- No Solution. The whole circle rotates towards
increasing angles.
By integrating the governing equation we get the
period to be:

, sinu e u a =

u e sin a =
0 , > a e
e < a
2 2
2
a
T

=
e
t
27
The non-uniform oscillator
, one fixed point ---> ,half-stable.



, there are two distinct fixed points, one stable
and one unstable. These are born out of a saddle-
node bifurcation at the critical value
e = a 2 * t u =
e < a
e = a
28
Two-dimensional linear systems
Ex: Spring mass system
29
Stability Language
If F.P attracts all trajectories in the phase plane
it is called global attracting.
If all trajectories that start sufficiently close to
F.P remains close to it for all time , the F.P is
Liapunov stable.
If nearby trajectories are neither attracted nor
repelled from F.P, then F.P is neutrally stable
(Center).

30
Classification of linear system





For the 2 x 2 matrix A
The characteristic equation:
Eigen-values:
Where:
31
Attracting directions
Trajectories approach the F.P tangent to slow
Eigen-direction , (Eigen-vector with the
smallest ||), and parallel to the fast Eigen-
direction.
32
Some special cases
2< 1 or 1< 2 : case (a) & (b).
1= 2= : Star node: (c)
1= 2= 0: infinite F.P,
1 or 2= 0: (d)
One Eigen-vector: Degenerate: (e)

0 = x

(a) (b) (c) (d) (e) 33

Classification of 2D linear system
Unstable Saddle point.
: Three cases:





0 < A
0 > A
0 4
2
> A t
0 < t
0 > t
Stable
Unstable
Node
0 4
2
= A t
Star
Degenerate
Node
Spiral
Center
0 4
2
< A t
0 < t
0 > t
Stable
Unstable
0 = t Neut. Stable
0 ] Re[ =
0 ] Re[ >
0 ] Re[ <
34
Classification of linear system
35
Nonlinear systems in the phase
plane
( )
( )

=
=
2 1 2 2
2 1 1 1
,
,
x x f x
x x f x

36
Linearization, Jacobean matrix
The stability of a fixed point is determined by
the Eigen-values and Eigen-vectors of the
Jacobean evaluated at the fixed point
|
|
|
|
.
|

\
|
c
c
c
c
c
c
c
c
=
2
2
1
2
2
1
1
1
x
f
x
f
x
f
x
f
J
) , ( *
*
2
*
1
x x x =
s Eigenvalue I J > = 0 | |
37
Classifications of the fixed points
Robust cases




Marginal cases
Unstable
Unstable
Stable
saddle
source
k
d distribute
repellers
attractors

+
>
< sin
,
0 ] Re[
0 ] Re[

Stable Center imaginary = 0 ] Re[

At least one is zero -----> high order & non-
isolated fixed point
38
Definitions
Hyperbolic F.P:
If
Structural stability phase portrait:
If the topology of the phase portrait cannot be
changed by arbitrary small perturbation to the vector
field. ( under damping Saddle is structurally stable ,
while center is not , convert to spiral).
Basin of attraction:
Its the set of ICs such that: x(t) x* as t .

0 ] Re[ =
s

39
Definitions
Basin boundary:
The line separates the basins for two nodes.
Separatix:
The line or orbit separates regions in the phase
space corresponds
to qualitatively different
dynamical behavior.
(separates stable and
unstable)
40
Conservative systems
This name is used for systems that originate from
Newton's law with a conservative force.


By integrating this system we will get the following :


Where:

is the kinetic energy .
V (x) is the potential energy.
E is the total energy.


dx
dV
x m =

E x V x m = + ) (
2
1
2

2
2
1
x m
41
Conservative systems
Rearrange the last equation to get the following
form:

The above eqn. can be used to plot the phase
portrait manually:
- Plot V(x) versus x.
- Plot several levels of E
on V(x) plot.
- Draw the difference (E-V)
on x-x plane
42
( ) ) (
2
x V E
m
x =
Conservative systems
If x* is a local minimum of V(x), then (x*, 0) is
a local minimum of E.
If x* is a local maximum of V(x), then (x*, 0) is
a saddle.
If (x*, y* = 0) is an isolated F.P of the
conservative system, and if its is local
minimum of E, then (x*, y *= 0) is a center
43
Pendulum
0 sin = + u u
L
g

u u
v
v
u
sin
L
g
= =
=

E
L
g
= u v cos
2
1
2
first-order form
Total energy
Fixed points
) 0 , (
) 0 , (
*
3
2
t
t
t u
=
=

=
P
P
] , [ t t u e
) 0 , 0 ( 0 *
1
= = P u
Jacobean
|
|
.
|

\
|

=
0 cos
1 0
u
L
g
J
44
Pendulum
Center L g i
L
g
J =
|
|
.
|

\
|

= /
0
1 0
) 0 , 0 (
2 , 1

Saddle L g
L
g
J =
|
|
.
|

\
|
= /
0
1 0
) 0 , (
2 , 1
t
45
Pendulums and their phase portraits.
46
Damped Pendulum
By adding damping the system now is not
conserved and from stability view Centers
become spiral and Saddle remains Saddle
47
Periodic Solution
Unlike equilibrium solution, its a dynamic
solution that characterized by time varying
states and one basic frequency.
For a continuous time system:
is a periodic solution with T is the
least period if:

) (t X x =
T
t X t X
t X T t X
< <
)
`

= +
= +
t
t
0
) ( ) (
) ( ) (
48
Autonomous System (A.S)


x: n-dimensional state vector.
M: m- dimensional parameter vector.
A periodic solution X with least T corresponds
to close orbit in the domain.
A periodic solution for A.S can be treated as a
F.P of a defined map called the Poincare map.
) ; ( M x F x =

49
Definitions in A.S
A periodic solution is called Limit cycle if there
are no other periodic solutions sufficiently
close to it.

Isolated: there are no other trajectories near
it, nearby trajectories either approach it
(stable) or go away from it (unstable).

50
Bendixsons criterion, 2D
For the A.S if

- Does not change sign, or does not vanish, then:
The periodic solutions are not possible in D.



No chance to change sign
Vanish for
( )
( )

=
=
2 1 2 2
2 1 1 1
,
,
x x f x
x x f x

2
2
1
1
x
f
x
f
c
c
+
c
c

+ =
=
2
3
1 1 2
2 1
2 x x x x
x x

2
2
2
1
1
=
c
c
+
c
c
x
f
x
f
Periodic = 0
51
Nonautonomus Systems (N.A.S)


A periodic solution X with least T corresponds
to close orbit in the domain.

We can see that the major difference between
the A.S and the N.A.S is that the last one
depends on time explicitly.

) ; , ( M t x F x =

52
Floquet Theory A.S
For the periodic solution of A.S:

is nxn constant matrix and called Monodromy
matrix. It maps an initial vector at t=0 to another
vector at t=T.

The Eigen-values () of the Mondromy matrix are
called Floquet multipliers. It will be used to classify
stability.

One of the Floquet multipliers of A.S is always Unity.


| ) ( ) ( t Y T t Y = +
|
53
A.S Stability
If only one =1 -----> Hyperbolic .
If more than one =1 -----> NonHyperbolic .
Hyperbolic:
- If all < 1 -----> Stable (Attractors).
- If one or more >1 -----> Unstable
- If all (other than unity) >1 -----> Unstable periodic
reppeller.
- If distributed ( <1 & >1) -----> Unstable limit cycle
of saddle type.



54
A.S Stability
Nonhyperbolic:

- If one or more >1 ----> Unstable.

- If all <1 -----> nonlinear analysis needed
55
N.A.S Stability
In contrast with the A.S case for which one of
the Floquet multipliers is always unity, in the
N.A.S such conditions is not satisfied.

If non lie on the unit circle ( 1) ----->
Hyperbolic .
Otherwise -----> NonHyperbolic.


56
N.A.S Stability
Hyperbolic:
- If all <1-----> Stable limit cycle.
- If one or more >1 -----> Unstable
- If all >1 -----> Unstable periodic reppeller.
- If distributed ( <1 & >1) unit circle ---->
Unstable saddle type.

NonHyperbolic: Nonlinear analysis is needed.

57
Poincare Maps
Poincare Section: Its a hypersurface in the state
space that is transverse to the flow of a given
system of equations.
A.S: N.A.S:



- n(x) is a vector normal to the section located at x.
- F(x) is the vector field describing the flow.

0 ) ( ) (
0 ) ( ). (
=
=
x F x n
or
x F x n
T
0 ) ; ( )] ( [ = t x F t x n
T
58
Definitions
Two sided section: section with different sign
for intersections.
One sided section: section with same sign for
all intersections.
Poincare Map: map of current intersection to
the subsequent intersection.

) (
1 m m
x P x =
+
59
Poincare Map N.A.S
Period (T) can be used to construct a Poincare
section.

To construct a Poincare section, we collect discrete
points at intervals of the period T.

If represents a point on a section, the Poincare
map P is defined by:
) , , ( ) (
0 0
t T t x P + = q q
60
Poincare Map N.A.S
In a neighborhood of the fixed point of the
map, we have:


Stability of the fixed point 0 can be obtained
by studying the Eigen-values of the Jacobian
matrix

|| || ) ( ) ( ) (
2
0 0 0
v O v P D P v P + + = + q q q
q
) (
0
q
q
P D
61
Stability analysis of N.A.S based on
If all 1 -----> Hyperbolic.
If one or more =1 --> NonHyperbolic.
Hyperbolic:
- If all <1 -----> Stable limit cycle.
- If all >1 -----> Unstable repellers
- If distributed ( <1 & >1) ----> Unstable
saddle type.
NonHyperbolic: Nonlinear analysis is needed.

) (
0
q
q
P D
62
Poincare Map A.S
The period associated with a periodic orbit is
not usually explicitly known.
Stability can be analyzed according to the
Jacobian matrix , elements on nth row and
nth column are zero except the diagonal
which is unity.
Stability from J matrix constructed from the
Jacobian matrix after deleting the nth row
and nth column.

P D
,
63
Stability analysis of N.A.S based on J
- If all <1-----> Stable limit cycle.

- If all >1 -----> Unstable repellers

- If distributed (<1 & >1) ----> Unstable saddle
type.
- If one or more =1, while the rest <1 ---->
Nonlinear analysis is needed.

64
Bifurcation of Periodic solution
Bifurcation occurs when Floquet multipliers leave the
unit circle through:
+1:
- Transcritical.
- Symmetry breaking
- Cyclic-fold
-1:
- Period-doupling
Real axis:
- Secondary Hopf
65
Symmetry-breaking Bifurcation
Break the symmetry of the periodic solution, similar
to pitchfork.
Supercritical (a): stable asymmetric coexist with
unstable symmetric on one side of the bifurcation
point.
Subcritical (b): unstable asymmetric coexist with
stable symmetric on one side of the bifurcation
point.

66
(a) (b)
Cyclic-fold
A branch of stable periodic solutions and a
branch of unstable periodic solutions coalesce
and obliterate each other at the bifurcation
point.
Similar to the saddle node bifurcation of the
fixed point.
67
Period-doubling Bifurcation
Branch of stable periodic solutions that exists
before the bifurcation continues as an
unstable branch of periodic solutions after the
bifurcation.
Supercritical: Branch of stable-doubled
periodic solutions created.
Subcritical: Branch of unstable-doubled
periodic solutions destroyed.

68
Period-doubling Bifurcation
In the subcritical case,
- unstable (2T) collides stable (T/2----> unstable
(lower period T/2).

After k successive period-doubling
bifurcations, we would have 2k points on the
corresponding Poincare section.
69
Transcritical Bifurcation
Stable and unstable periodic solutions meet
and exchange their stability after bifurcation
point.

Similar to Transcritical in F.P.
70
Secondary Hopf or Neimark
Stable prior to the bifurcation continues as
unstable after the bifurcation.

Supercritical: A branch of stable quasiperiodic
solutions is created.
Subcritical: A branch of stable quasiperiodic
solutions is destroyed.

71
Thank You
72