Assignment I

03313104 Nonlinear Systems

and Control

Work Submitted by:

Josue Manuel Pareja Contreras
LS1925202

International School
北京航空航天大学
November 11, 2020
1 For following systems, identify type of the equilibrium point.
1. ÿ + y = εẏ(1 − y − ẏ 2 )
Obtaining System’s equations in term of X1 and X2 :

f1 = x˙1 = x2
f2 = x˙2 = εx2 − εx2 x1 − εx32 − x1

Finding the equilibrium point (EP), for this x˙1 = 0 and x˙2 = 0:

x˙1 = x2 = 0
x˙2 = x1 = 0

As a consequence equilibrium (EP) point is located at the origin x = (0, 0).

Now we proceed to linearize around the EP.

 
∂f1 ∂f1
∂f ∂x1 ∂x2
A= =
 
x1 = 0
∂x

x2 = 0 ∂f2 ∂f2 x1 = 0
∂x1 ∂x2 x2 = 0

From previous equation we obtain:

 
0 1
A=
−1 ε

Finding the Eigen Values of A matrix:

−λ 1
det(A − λI) =
−1 ε−λ

From previous determinant calculation we obtain the equations:

P
tr(A) = i λi
ε = λ1 + λ2

and:
det(A) = λ1 λ2 ...λn
1 = λ1 λ2

1
 From determinant equation it can be inferred that λ1 and λ2 must be both
positive or negative. For system’s stability we need both eigen to be less
than zero:

λ1 < 0 λ2 < 0

∴ From trace equation we can inferred that ε must be negative (ε < 0)

for the system’s equilibrium point at (0, 0) to be a stable equilibrium
point. If ε is positive (ε > 0) the equilibrium point is unstable
2. x˙1 = x2
x˙2 = −x1 + x2 (2 − 3x21 − 2x22 )
Finding the equilibrium point:

x˙1 = x2 = 0
x˙2 = x1 = 0

As a consequence equilibrium (EP) point is located at the origin x = (0, 0).

Now we proceed to linearize around the EP.

 
∂f1 ∂f1
∂f ∂x1 ∂x2
A= =
 
x1 = 0
∂x

x2 = 0 ∂f2 ∂f2 x1 = 0
∂x1 ∂x2 x2 = 0

From previous equation we obtain:

 
0 1
A=
−1 2

Eigen values from A matrix are obtained:

−λ 1
det(A − λI) =
−1 2−λ

Calculating Eigen values:

det(A − λI) = (−λ)(2 − λ) + 1 = 0

−2λ + λ2 + 1 = 0
(λ − 1)2 = 0
λ1,2 = 1

λ1,2 > 0

∴ The equilibrium point at x = (0, 0) is a unstable node.

2
3. x˙1 = x2
x˙2 = −x1 + x2 − 2(x21 + 2x2 )x22
Finding the equilibrium point (EP), for this x˙1 = 0 and x˙2 = 0:

x˙1 = x2 = 0
x˙2 = x1 = 0

As a consequence equilibrium (EP) point is located at the origin x = (0, 0).

Now we proceed to linearize around the EP.

 
∂f1 ∂f1
∂f ∂x1 ∂x2
A= =
 
x1 = 0
∂x

x2 = 0 ∂f2 ∂f2 x1 = 0
∂x1 ∂x2 x2 = 0

From previous equation we obtain:

 
0 1
A=
−1 1

Eigen values from A matrix are obtained:

−λ 1
det(A − λI) =
−1 1−λ

Calculating Eigen values:

det(A − λI) = (−λ)(1 − λ) + 1 = 0

λ2 − λ + 1 =√0
λ1 = 0.5 + 0.5√3i
λ2 = 0.5 − 0.5 3i

real(λ1 ) > 0
real(λ2 ) > 0

∴ The equilibrium point at x = (0, 0) is unstable focus.

3
4. x˙1 = x1 + x2 − x1 h(x)
x˙2 = −2x1 + x2 − x2 h(x), with h(x) = max{|x1 ||, |x2 |}
Finding the equilibrium point:

x˙2 = 0 = −2x1 + x2 − x2 h(x)

0 = −2x1 + x2 (1 − h(x))
x1 = 21 (x2 (1 − h(x)))

x˙1 = 0 = x1 + x2 − x1 h(x)
0 = x2 + x1 (1 − h(x))
0 = x2 + x2 21 (1 − h(x))2
0 = x2 (1 + 12 (1 − h(x))2 )
x2 = 0
∴ x1 = 0

The equilibrium (EP) point is located at the origin x = (0, 0). Now we
proceed to linearize around the EP.

 
∂f1 ∂f1
∂f ∂x1 ∂x2
A= =
 
x1 = 0
∂x

x2 = 0 ∂f2 ∂f2 x1 = 0
∂x1 ∂x2 x2 = 0

From previous equation we obtain:

 
1 1
A=
−2 1

Eigen values from A matrix are obtained:

1−λ 1
det(A − λI) =
−2 1−λ

Calculating Eigen values:

det(A − λI) = (1 − λ)2 + 2 = 0

1 − 2λ + λ2 + 2 = 0
λ2 − 2λ + 3√= 0
λ1 = 1 + √2i
λ2 = 1 − 2i

real(λ1 ) > 0
real(λ2 ) > 0
∴ The equilibrium point at x = (0, 0) is unstable focus.

4
2 Determine if following systems are stable, asymptotically stable
or unstable using Lyapunov’s Method.
1. x˙1 = sin(x2 )
x˙2 = 0
On Figure 1 the system’s phase portrait can be observed. The red asterisk
dots indicate the equilibrium points.

Figure 1: Phase portrait for exercise 2.1

Finding the equilibrium point:

x˙1 = 0 = sin(x2 )
x2 = asin(0)

∴ x2 = nπ, n ∈ W
x1 is free

Next Lyapunov candidate is proposed:

V = (1 − cos(x2 ))

5
 Note that if we analyze the value of V at (0,0) we obtain:

V (0, 0) = 0
From the value of V at any other point we obtain that:

V > 0, ∀x 6= 0
∴ V is positive definite. Now we proceed to analyze V̇ :

V = (1 − cos(x2 ))
V̇ = sin(x2 )x˙2
(1)
V̇ = sin(x2 ).0
V̇ = 0

Since V̇ is equal to 0 we can say that the system is just stable.

2. x˙1 = sin(x2 )
x˙2 = −x1
From phase portrait observed at Figure 2 we can see that the system
behaves like a stable system, due to the fact that the system does not
converge to the equilibrium point.

Figure 2: Phase portrait for exercise 2.2

6
Finding the equilibrium points:

x˙1 = 0 = sin(x2 )
x2 = asin(0)
x2 = nπ, n ∈ W

x˙2 = 0 = −x1
x1 = 0

The system’s equilibrium points are located at (0, nπ). The next Lyapunov
candidate is proposed:

V = 21 x21 + (1 − cos(x2 ))
V (0, 0) = 0
V > 0, ∀X 6= 0

∴ the proposed Lyapunov candidate is positive definite. Now we proceed

to analyze V̇ :

V = 21 x21 + (1 − cos(x2 ))
V̇ = x1 x˙1 + sin(x2 )x˙2
V̇ = x1 sin(x2 ) − x1 sin(x2 ) = 0

Since V̇ is equal to 0 we can say that the system is just stable.

7
3. x˙1 = sin(x2 )
x˙2 = −x1 − x2
From phase portrait observed at Figure 3 we can see that the system
behaves like an asymptotic stable system for some equilibrium points, due
to the fact that the system converges to the stable equilibrium point and
diverges from the unstable ones.

Figure 3: Phase portrait for exercise 2.3

Finding the equilibrium points:

x˙1 = 0 = sin(x2 )
x2 = asin(0)
x2 = nπ, n ∈ W

x˙2 = 0 = −x1 − x2
x1 = −x2
x1 = −nπ, n ∈ W

8
The system’s equilibrium points are located at (−nπ, nπ). A proper Lya-
punov candidate must be proposed so from it asymptotic stability can
be proved. The Lyapunov candidate from previous exercise 2.2 can only
assure simple stability. Now we propose a new Lyapunov candidate:

V = X T P X + (1 − cos(x2 ))
  
  P11 P12 x1
V = x1 x2 + (1 − cos(x2 )
P12 P22 x2

V = P11 x21 + 2P12 x1 x2 + P 22x22 + (1 − cos(x2 ))

V (0, 0) = 0
V > 0, ∀X 6= 0

Where P is a positive define symmetric matrix. Now we proceed to analyze

de derivative V̇ .

V̇ = (P11 − 1)x1 sin(x2 ) + (P12 − 1)x2 sin(x2 ) − x1 x2 (P12 + P22 ) − P12 x21 − P22 x22

First, for V̇ to be negative definite it needs to satisfy the conditions:

0 < P12 < 1

P22 > 0

Second, from this two conditions we can assume P12 = P22 = 0.5. Next
condition should be also satisfied:

(P11 − 1)x1 sin(x2 ) − x1 x2 (P12 + P22 ) < 0

Which can be simplified to:

(P11 − 1)sin(x2 ) − x2 < 0

(P11 − 1)sin(x2 ) < x2

It can be seen on Figure 4 that |x2 | > |sin(x2 )|, ∀x2 .

9
 Figure 4: y = |x2 | vs y = |sin(x2 )|

We can see that close to the origin |x2 | ≈ |sin(x2 )|, therefore:

(P11 − 1)|x2 | − |x2 | < 0

P11 − 1 − 1 < 0
P11 < 2

To keep simplicity we can choose P11 = 0.5. As a result our P matrix has
the form:
 
0.5 0.5
P =
0.5 0.5

Finally our Lyapunov candidate can be rewritten as:

V = X T P X + (1 − cos(x2 ))
  
  0.5 0.5 x1
V = x1 x2 + (1 − cos(x2 )
0.5 0.5 x2

V = 21 (x1 + x2 )2 + (1 − cos(x2 ))

V (0, 0) = 0
V > 0, ∀X 6= 0

10
Where V̇ is equal to:

V̇ = x1 x˙1 + x˙1 x2 + x1 x˙2 + x2 x˙2 + x˙2 sin(x2 )

V̇ = x1 sin(x2 ) + x2 sin(x2 ) + x1 (−x1 − x2 ) + x2 (−x1 − x2 ) + sin(x2 )(−x1 − x2 )
V̇ = −(x21 + 2x1 x2 + x22 )
V̇ = −(x1 + x2 )2
V̇ (x1 , x2 ) < 0, ∀x 6= 0

Since V̇ (x1 , x2 ) is negative definite and V (x1 , x2 ) is positive definite we

can conclude that the system is Asymptotically stable.

11
4. x˙1 = sin(x2 )
x˙2 = −x1 + x2
From phase portrait on Figure 5 we can see that all system’s equilibrium
points are unstable, since trajectories diverge to infinity instead of con-
verging to 0 or any other equilibrium point.

Figure 5: Phase portrait for exercise 2.4

Finding the equilibrium points:

x˙1 = 0 = sin(x2 )
x2 = asin(0)
x2 = nπ, n ∈ W

x˙2 = 0 = −x1 + x2
x1 = x2
x1 = nπ, n ∈ W

The system’s equilibrium point are located at (nπ, nπ). A similar Lya-
punov candidate to the one obtained on exercise 2.3 is used:

12
 V = 21 (x1 − x2 )2 + (1 − cos(x2 ))
V (0, 0) = 0 (2)
V > 0, ∀X 6= 0

As a consequence we can say that V is positive definite. Now we can

proceed to analyze V̇ .

V̇ = x1 x˙1 − x˙1 x2 − x1 x˙2 + x2 x˙2 + x˙2 sin(x2 )

V̇ = x1 sin(x2 ) − x2 sin(x2 ) − x1 (−x1 + x2 ) + x2 (−x1 + x2 ) + sin(x2 )(−x1 + x2 )
V̇ = x21 − 2x1 x2 + x22
V̇ = (x1 − x2 )2
V̇ (x1 , x2 ) > 0, ∀x 6= 0

Since V̇ (x1 , x2 ) is also positive definite we can conclude that the system
is unstable.

13
3 Prove that the origin is asymptotically stable in the following
system, and estimate the region of attraction
x˙1 = −x1 + x2
x˙2 = −x2 − x1 + x22
Finding the Equilibrium Point

x˙1 = 0 = −x1 + x2
x1 = x2

x˙2 = 0 = −x2 − x1 + x22

−2x2 + x22 = 0
x2 (x2 − 2) = 0
∴ x2 = 0, x2 = 2
The system has two equilibrium points, one located at the origin (0,0) and
the other at (2,2). Sin the problem only asks for the stability characteristics of
the origin we will only focus on this point, the other equilibrium point will be
ignored.
Now we proceed to linearize around the equilibrium point.

 
∂f1 ∂f1
∂f ∂x1 ∂x2
A= =
 
x1 = 0
∂x

x2 = 0 ∂f2 ∂f2 x1 = 0
∂x1 ∂x2 x2 = 0

From previous equation we obtain:

 
−1 1
A=
−1 −1
Eigen values from A matrix are obtained:

−1 − λ 1
det(A − λI) =
−1 −1 − λ
Calculating Eigen values:

det(A − λI) = (−1 − λ)2 + 1 = 0

1 + 2λ + λ2 + 1 = 0
λ2 + 2λ + 2 = 0
λ1 = −1 + i
λ2 = −1 − i
real(λ1 ) < 0
real(λ2 ) < 0
∴ The equilibrium point is a Stable Focus. Now lets calculate the Lyapunov
function by using the Lyapunov Equation 3 from the linearized system around
the origin:

14
 P A + AT P = Q (3)
Where Q is a symmetric positive definite 2x2 matrix that due to simplicity
we will keep as the identity matrix:
 
1 0
Q=
0 1
Some lines of code in Octave/Matlab can give us the P matrix as a result:
1 A = [ -1 1; -1 -1];
2 Q = eye (2 ,2) ;
3 P = lyap (A ' , Q )

The obtained value of P matrix is:

 
0.5 0
P =
0 0.5
Next, the Lyapunov Function and its derivative for the linearize system
around the equilibrium point at the origin is obtained:

V = XT P X
V = 12 x21 + 12 x22
The Lyapunov function is positive definite for all values of x except at the
origin where is 0. Now we proceed to analyze its time derivative:

V̇ = −x21 − x22 + x32

V̇ = −x21 + (x2 − 1)x22
To guarantee Asymptotic Stability we need:

x2 − 1 < 0
x2 < 1
So:

V̇ < 0, ∀|x2 | < 1

The bound |x2 | < 1 can be expressed as:

|x2 | < 1
 
  x1
0 1 <1
x2

|bT x| < r

15
From previous equations we can obtain:
 
0
b= , r=1
1
The system’s region of attraction is defined as:

Ωc = {V (x) < c}
Where c is obtained from:
r
c<
bT P b
1
c<   
  0.5 0 0
0 1
0 0.5 1

c < 1/0.5
c<2
Region of attraction Ωc is defined as:

Ωc = {V (x) < 2}

16
4 Consider system
x˙1 = −x1
x˙2 = (x1 x2 − 1)x32 + (x1 x2 − 1 + x21 )x2

ˆ Prove that the origin is the sole equilibrium point:

Finding the equilibrium point of the system:

x˙1 = 0 = x1

x˙2 = 0 = (0 − 1)x32 + (0 − 1 + 0)x2

0 = −x3 − x2
x2 (x2 +2 +1) = 0

We can see that the resultant polynomial has three roots, two of them
obtained from (x2 +2 +1) are imaginary. On the other hand the third root
X2 = 0 can be the only valid X2 component of the system’s equilibrium
point.
∴ The origin is the sole equilibrium point

ˆ Prove that the origin is asymptotically stable through linearization

 
∂f1 ∂f1
∂f  ∂x1 ∂x2
A= =

x1 = 0
∂x

x2 = 0 ∂f2 ∂f2 x1 = 0
∂x1 ∂x2 x2 = 0

From previous equation we obtain:

 
−1 0
A=
0 −1

Eigen values from A matrix are obtained:

−1 − λ 0
det(A − λI) =
0 −1 − λ

Since A is diagonal matrix the eigen values are equal to the diagonal
elements:

λ1,2 = −1 (4)

The equilibrium point at the origin is asymptotically stable due to the fact
that Eigen values of matrix A show that the origin is a Stable Node.

17
ˆ Prove Γ = {x ∈ R2 |x1 x2 ≥ 2} is a positive invariant set. To prove that Γ
is a positive invariant set we let :

V = x1 x2

Time derivative of V L

V̇ = x˙1 x2 + x1 x˙2
V̇ = −x1 x2 + x1 (x1 x2 − 1)x32 + x1 (x1 x2 − 1 + x21 )x2
V̇ = −2 + 2x22 + 2 + 2x21
V̇ = 4x22

The value of V̇ is strictly positive, ∴ Γ is a positive invariant set.

ˆ Is the origin globally asymptotically stable?
No, since Γ has been proved to be a positive invariant set, we can say that
all trajectories starting from it diverge to infinity instead of converging to
the origin.

18