CL 714: Nonlinear System Analysis

Definitions-1
1. System: System is a set of interacting or interdependent components forming an integrated
whole.
2. State space of system: To study any system we identify some of its attributes which are of
interest to us. The set

of the variables which quantify all the attributes of interest is called

the state space of the system. State space is also called the phase space. For example, in
statistical mechanics, the state space of the system involving
position coordinates and

particles, consists of

components of momenta.

3. State of system: A point

is called the state of the system.

4. Dynamical system: It is the system, the state of which evolves in time, i.e.

( ). We

will define it more precisely later.

5. Process: Process is a set of rules which governs the evolution of the state of the system.
The process governing the system can be physical, chemical, biological, ecological,
economical, social or combination of two or more of these. The process is called
deterministic, if the governing laws are precisely known. Thus given the initial state of the
system and the process governing the system are specified, future state of the system can be
precisely predicted as a function of time. It is called the stochastic system if the laws are not
precise and are governed by probability. In the present course, we shall consider only
deterministic systems.
6. Two types of dynamical systems: The time belongs to set , i.e.
continuous i.e.

or it could be discrete , i.e.

, The set

can be

. The former system is called the

continuous time dynamical system and the latter is called the discrete time dynamical system
(state in discrete time dynamical system is denoted by

.)

7. Evolution operator: The rules governing the evolution of the system can be embedded in
. It maps the initial state ( ) to state

the evolution operator, which is denoted by

( ), which corresponds to time .
( )

( )

For example for a system governed by a differential equation

, the evolution

equation is
( )

( )

The evolution operator is

. Note that it is not always possible to express the evolution

operator explicitly. However, it exists for every dynamical system.

Evolution operator for a deterministic system must obey two important characteristics. First
is that

is an identity operator i.e.

( )

( )

for all ( )

and

Second property is given by

(

or

) for all

and

Evolution operator
Thus for the previous example we have
(

( ( )

We call evolution operator invertible if, it exists also for negative times i.e.
that .

exists such

is an identity operator. In the example above, the evolution operator

invertible and its inverse is

is

For a discrete time system, evolution operator is fully specified by defining

, called

time-one map . Thus

is generally denoted by

to distinguish it from continuous operator .

From the property of evolution operator we have

and so on. For example, for a system governed by difference equation

time map is

, which gives

8. Example: Find the evolution operator

and

, the one

for the equation

and verify its basic properties

Solution: Integration yields

If at

or

, then

or

or

or
( )

From this we obtain

Note that
( )

Hence

is an identity operator. Similarly

( )
and

( )

)
) (

8. Example: Find the evolution operator

( )

for the set of equation

and

Solution: These equations can be solved to yield

and

The solution can be written as

( )
]
( )

+[

( )
] , Hence
( )

+, where

9. Dynamical system (definition): It is a triple*

space and

is a time set,

is a family of evolution operators, parameterized by

is a state
and

satisfying the properties described earlier.

10. Solution curve: If we plot the state of a system as a function of time, beginning from
predefined initial state

(to which we assign

) the resulting curve is called the solution

curve starting from

( )

*(

) (

acts as a parameter on the solution curve. Solution curve is obtained by operating

(

on

11. Dynamical systems governed by differential equations: In this course, we are

particularly interested in the systems which are governed by differential equations written in
the form:

dx
X x , a , t
dt
In a more expanded form it can be written as
dx1
dt X 1 x1 , x2 ,, xn , a1 , a2 ,, am , t
dx

2 X 2 x1 , x2 ,, xn , a1 , a2 ,, am , t
dt

dxn X n x1 , x2 ,, xn , a1 , a2 ,, am , t
dt

X Set of functions, x set of dependent variables (state variables), t independent

variable, a Set of parameters (these are treated as constants while solving the equations but
treated as variables while studying their effect on the solution).
The set

x x0 at t t0
forms the set of initial conditions.
The set of functions X can be separated into state functions and input functions
(also called forcing functions, stimuli). The state functions are determined by the nature of
the system. The input functions can be externally altered.
If one or more of the functions X1, X 2 ,, X n , are nonlinear in x1 , x2 ,, xn , then the
set of the equations is called the nonlinear set of ODEs.
12. Example: As an example consider a linear system whose governing equation is

d 2x
dx
kx f t
2
dt
dt

It is a toy model for a mechanical system in which the first term represents product of mass
and acceleration ( x displacement). Second term corresponds to a friction force ( damping
force), the third term to the restoring force and the fourth term to the forcing function, which
may corresponds to the external force. The equation can be cast as set of two ODEs by setting
x1 x and x2

dx
. The resulting ODEs are
dt

dx1
x2
dt
dx2

k
1
x2 x1 f t
dt
m
m
m
In the matrix form, we write

dx1
1 x1 0
dt 0
dx
x f t m (1)

k
m

m
2

dt
x
0
Here 1 is the vector representing the state of the system,
is the vector
f t m
x2

representing the forcing functions. k m and m are the parameters.

If the restoring force varies as kx12 , we get a nonlinear system

dx1
1 x12 0
dt 0

dx
f t m (2)

k
m

m
2

x2

dt
If we are not interested in the effect of parameters, then we can simplify the notation
and write the set as

dx
X x , t
dt
In these equations parameters would appear as constant coefficients in the terms on the right.
13. Autonomous and non-autonomous systems: If t does not appear explicitly in X , then
we can write

dx
X x
dt

The above set of equation is called the autonomous system. If t appears explicitly in the
set, then the set forms nonautonomous system.
Autonomous system possesses translational symmetry, that is, if x t is the solution
of

dx
X x , with the initial condition, x x0 at t t0 , then x t c is the solution of
dt

dx
X x with x x0 at t t0 c . To prove this, let us substitute t c in the shifted
dt
differential equation to obtain

dx
dx
X x or
X x
d c
d
The initial condition x x0 at t t0 c reduces to

x x0 at t 0
The solution of the resulting differential equation is x x x t c as expected.
This translational symmetry is illustrated in the figure below. It is important to notice
that the shapes of the original and the translated curves are the same since shape is
determined by X x which does not contain t . Hence it is affected by x0 , but not t0 .

14. Phase space: It is a geometric representation of state space of a system. Suppose a system
has n-dimensions, then each state can be represented by a point

in n-dimensional space.

Two dimensional phase space is also called phase plane. Note that each point in phase space
represents state of the system corresponding to time
(

and some initial condition

, i.e.

) . However, in phase space representation, we neither mention time, nor the initial

condition, explicitly. Hence this representation is not useful unless we associate some more
information with this representation.
15. Vector field: Dynamical system is characterized by the fact that its state changes with
time. Hence, the point representing its state in state space is, in general, moves with time. The
time derivative

represents the velocity of the point

in the state space. This velocity can

be represented by an arrow in the phase space. The absolute value of velocity determines the
length of the arrow and the direction in which the arrow points depicts the direction in which
the particular state is moving in the state space. There is an arrow associated with every point
in the state space. Set of all these arrows constitute the vector field associated with the given
dynamical system. It is cumbersome to represents the length of arrow and in qualitative
analysis we do not need the magnitude of velocities. Hence in many cases lengths of the
arrows are arbitrary. If these arrows are generated by computer software, then the arrows are
coloured using a colour scheme which is associated with the magnitude of velocity. An
example is shown in the figure below.

x1 x1 and x2 2x2

16. Isoclines: The level curves

are called the isoclines. Different isoclines are

obtained for different values of . The velocity vector at each point on the isocline is
identical.
17. Example: Sketch isoclines for the differential equation

Solution: Isoclines are obtained by solving

for different values of .

Rearrangement gives

Isoclines are the plots of

versus with as a parameter. The plots for six different values

of the parameter are as shown in the figure below.

10

20

20

10

10

10

Isoclines for

18. Orbit or trajectory: It is the basic geometrical object associated with a dynamical
system *

+. It is defined as an ordered subset of the state space of the system. The

ordering is done by time.

Orbit is obtained by projecting the solution curve on to the phase space

A typical orbit is shown below. The arrow indicates the direction of increasing time.

Note that an orbit is a curve having same dimension as the state space. Also time is not
explicitly displayed on orbit as illustrated in the example below.
19. Example: Find the equation of the orbit passing through point (1, 1) at

of the

dynamical system

and

Solution: The solution of the equations is

and
From the condition

and

at

we obtain

, hence

and
The equation of the orbit is obtained by eliminating time from these equations

20. Map
Discrete system can be represented graphically as a map (also called cobweb) as shown in
the figure below

x2

x1
xn1 f xn

x0

x1

x2

xn1 xn

Here, the abscissa represents xn and the ordinate xn1 . The diagonal line is xn1 xn and the
curve represents xn1 f xn . Staring from x0 , we follow the staircase as shown in the
figure.
21. Fixed point (equilibrium point): A point
point if

for all

is called the fixed point or equilibrium

. For example, the fixed points for a system governed by

differential equation

dx
X x , t
dt
are obtained by solving the set of equations

X x , t 0
Note that for nonautonomous systems, fixed point may change with time. For autonomous
systems, they are independent of time.
For discrete time dynamical system defined by equation
(

The fixed pints are obtained by solving the equation

( )

22. Example: Find the fixed points of

)(

Solution: The fixed points are obtained by solving

(
The fixed points are:

and

)(

23. Example: Find the fixed points for the discrete system

10

The fixed points are obtained by solving the following

or (

and

24. Cycle: A cycle

for some

are the two fixed points.

is an orbit satisfying the condition

, for all

for every

is called the period of the cycle. Cycle is, therefore a

periodic orbit or a closed orbit.

The figure below shows two cycles, cycle (a) is for a continuous time system and (b)
is for a discrete time system.

For a discrete time system cycle represents a finite and fixed set of points.
( )
(Note that

( )

( )

is the fixed point of the evolution operator

( ).)

25. Example: Show that the orbits of the dynamical systems governed by the following
equation x1 x2 and x2 x1 are cycles. Find the direction of cycles and their period.
Solution: (a) We can obtain the trajectories by two methods. In the first, we obtain the
evolution operator. Next we plot trajectories in three dimensions(
the trajectories on the plane (

). Next we project

). A simpler procedure is to time by dividing the first

equation by the second. This yields the following equation

dx2
x
1
dx1
x2

or

x12 x22 c 2

The trajectories form a family of circles with centre as origin and radius c . They are shown
in the figure blow. Note the clockwise direction of the arrow, which represents flow of time.
To find this direction, we proceed as follows. From the original equations, we see that in the
first quadrant,

, but

. This represents clockwise flow of trajectories.

11

In order to find the period we use the following transformations

x1 r cos and x2 r sin
Differentiation with respect to time yields

x1 r cos r sin

and

x 2 r sin r cos

Substitution into the original equation yields

r cos r sin r sin and r sin r cos r cos

Solving these equations, we get

r 0

and 1

The first equation predicts that the cycle is a circle. Second equation says that the angular
velocity of point on the circle is 1 . Hence it will take time equal to 2 in order to complete
one cycle. Hence the period of the cycle is 2 and is independent of the radius.
26. Limit cycle: A cycle of a continuous-time dynamical system, in the neighbourhood of
which there are no other cycles is called a limit cycle. Limit cycle is, thus, an isolated cycle.
The circles in the previous example are not limit cycles, since in the neighbourhood of
every cycle there is another cycle. Next example illustrates a limit cycle.
27. Example: Show that the orbits of the dynamical systems governed by the following
equation

x1 x2 , x2 x12 x22 1 x2 x1

has a limit cycle, which is a circle x12 x22 1 .

Solution: Using the transformation, x1 r cos and x2 r sin , we can convert these
equations to
12

r cos r sin r sin and r sin r cos r 2 1 r sin r cos

In the matrix form, we can write

* + *

Inverting the equations we get

* +

Note that when

( )

+[

(
(

shows that

, we get r 0 and 1 which describes the circle of unit radius. If

, r 0 and radius of a trajectory will decrease till it reaches

hand if ( )

. On the other

, r 0 and radius of a trajectory will increase till it reaches

. This

is a limit cycle since is a cycle and there are no cycles in its neighbourhood.

13