1.

For a dynamical system involving n state variables (the state vector is 𝑥), we can choose
𝑥 𝑇 𝐴𝑥 (where 𝐴 is a matrix of dimension 𝑛 × 𝑛) as a valid Lyapunov function if : (MULTIPLE
SELECT QUESTION)
A) principal determinants of 𝐴 − 𝐴𝑇 are positive
B) principal determinants of 𝐴 − 𝐴𝑇 are negative
C) principal determinants of 𝑨 + 𝑨𝑻 are positive
D) principal determinants of 𝐴 + 𝐴𝑇 are negative
E) principal determinants of 𝐴𝐴𝑇 are positive

F) principal determinants of 𝐴𝐴𝑇 are negative

G) none of these

Solution :

𝑉(𝑥) = 𝑥 𝑇 𝐴𝑥 will be a valid Lyapunov function, if it will be positive definite.

As 𝑉(𝑥) = 𝑥 𝑇 𝐴𝑥 is a scalar , so we can write

𝑇
(𝑉(𝑥)) = 𝑉(𝑥)

=> 𝑥 𝑇 𝐴𝑇 𝑥 = 𝑥 𝑇 𝐴𝑥
So we can write
𝑇
(𝑉(𝑥)) + 𝑉(𝑥) 𝑇
𝐴𝑇 + 𝐴
𝑉(𝑥) = =𝑥 ( )𝑥
2 2

As 𝐴𝑇 + 𝐴 is a symmetric matrix, if the principal determinants of 𝐴𝑇 + 𝐴 will be positive, then it

will be a positive definite matrix. Hence option C is correct.

2. For a dynamical system involving the states as 𝑥1 , 𝑥2 , 𝑥3 , the function 𝜙(𝑥) = 𝑥12 + 4𝑥22 +
𝑥32 + 3𝑥1 𝑥2 + 4𝑥2 𝑥3 + 3𝑥1 𝑥3 is :
A) Positive definite
B) Positive semi-definite
C) Negative definite
D) Negative semi-definite
E) None of these (sign indefinite)

Solution :

Here the function 𝜙(𝑥) can be written as :

𝜙(𝑥) = 𝑥 𝑇 𝐴𝑥
1 3 3
With 𝐴 = [0 4 4]
0 0 1
𝐴𝑇 +𝐴 1
𝜙(𝑥) can be written as 𝜙(𝑥) = 𝑥 𝑇 ( 2
) 𝑥; = 2 𝑥 𝑇 𝑄𝑥

2 3 3
=> 𝑄 = 𝐴 + 𝐴𝑇 = [3 8 4]
3 4 2
 Now the principal determinants can be calculated as :

det (2) = 2
2 3
det ( )=7
3 8
2 3 3
𝑑𝑒𝑡 (3 8 4) = −18
3 4 2
Two of the principal determinants of 𝑄 are positive, and one is negative. So 𝑄 is sign indefinite.

Sign definiteness of a symmetric matrix can also be found using eigenvalue analysis. For positive
definiteness the requirement is to have all the eigen values positive. For negative definiteness the
requirement is to have all the eigen values negative. If some are positive and some are negative, then
sign indefinite.

3. Consider the dynamical system

𝑥̇ 1 = −6𝑥1 − 12𝑥2 : = 𝑓1
𝑥̇ 2 = −2𝑥1 − 5𝑥2 − 𝑥23 ≔ 𝑓2
Define 𝑓 = [𝑓1 ; 𝑓2 ]. If jacobian of 𝑓 with respect to [𝑥1 𝑥2 ]𝑇 is denoted as 𝐴, then using
Kravoskii’s Theorem, we can conclude that : (MULTIPLE SELECT QUESTION)

A) The system is asymptotically stable, because all the principal determinants of (−𝐴 − 𝐴𝑇 )
are positive ∀𝑥1 , 𝑥2 ∈ 𝑅.
B) Asymptotic stability can’t be guarenteed, because all the principal determinants of
(−𝑨 − 𝑨𝑻 ) are not positive ∀𝒙𝟏 , 𝒙𝟐 ∈ 𝑹.
C) The system is globally asymptotically stable, because all the principal determinants of
(−𝐴 − 𝐴𝑇 ) are positive ∀𝑥1 , 𝑥2 ∈ 𝑅 and 𝑉 = 𝑓12 + 𝑓22 is radially unbounded (meaning :
𝑓12 + 𝑓22 = ∞).
D) None of these.

Solution :

The jacobian is :
−6 −12 6 12
𝐴=[ ] = −[ ]
−2 −5 − 3𝑥22 2 5 + 3𝑥22
12 14
We have to check sign-definiteness of 𝑄 = 𝐴 + 𝐴𝑇 = − [ ]
14 10 + 6𝑥22
12 14
At 𝑥2 = 0, 𝑄 = − [ ]
14 10
The principal determinants of −𝑄 are 12 and -76. So we can not conclude asymptotic stability by
Kravoskii’s theorem. However for 𝑥22 > 76/72 , 𝑄 is negative definite. So the system is stable in the
sense of Lyapunov, but asymptotic stability can’t be concluded.

4. Consider the dynamical system

𝑥̇ 1 = −6𝑥1 + 2𝑥2 : = 𝑓1
𝑥̇ 2 = 2𝑥1 − 5𝑥2 − 𝑥23 ≔ 𝑓2
 Define 𝑓 = [𝑓1 ; 𝑓2 ]. If jacobian of 𝑓 with respect to [𝑥1 𝑥2 ]𝑇 is denoted as 𝐴, then using
Kravoskii’s Theorem, we can conclude that : (MULTIPLE SELECT QUESTION)

A) The system is asymptotically stable because all the principal determinants of (−𝑨 −
𝑨𝑻 ) are positive ∀𝒙𝟏 , 𝒙𝟐 ∈ 𝑹.
B) The system stability can’t be guaranteed, because all the principal determinants of
(−𝐴 − 𝐴𝑇 ) are not positive ∀𝑥1 , 𝑥2 ∈ 𝑅.
C) The system is globally asymptotically stable, because all the principal determinants of
(−𝑨 − 𝑨𝑻 ) are positive ∀𝒙𝟏 , 𝒙𝟐 ∈ 𝑹 and 𝑽 = 𝒇𝟐𝟏 + 𝒇𝟐𝟐 is radially unbounded (meaning :
𝒇𝟐𝟏 + 𝒇𝟐𝟐 = ∞).
D) None of these.

Solution :

The jacobian is :
−6 2 6 −2
𝐴=[ ] = −[ ]
2 −5 − 3𝑥22 −2 5 + 3𝑥22

The principal determinants of −(𝐴 + 𝐴𝑇 ) are 12 and 136 + 72𝑥22 , which are positive for
all values of 𝑥1 and 𝑥2 . So the system is asymptotically stable.

Further 𝑉 = 𝑓12 + 𝑓22 = (6𝑥1 − 2𝑥2 )2 + (2𝑥1 − 5𝑥2 − 𝑥23 )2 is positive definite and radially
unbounded. So the system is globally asymptotically stable.

Tips for Q5,Q6,Q7,Q8

If for a system 𝑥̇ = 𝑓(𝑥), 𝑥 ∈ 𝑅 𝑛 (with 𝑥 = 0𝑛×1 as the only equilibrium point) we take a radially
unbounded positive definite Lyapunov candidate function as 𝑉(𝑥) and by differentiating with respect
to time, we get :

(i) 𝑉̇ < 0 everywhere except at origin, then the system is globally asymptotically
stable.
(ii) 𝑉̇ < 0 everywhere except in a closed set 𝒮 containing the origin (but the
boundary of the set doesn’t contain the origin), then the system is stable in the
sense of Lyapunov, but asymptotic stability can not be concluded using the
Lyapunov function 𝑉(𝑥), another Lyapunov function may be taken for further
analysis.
(iii) 𝑉̇ < 0 inside the set 𝒮 except at origin and boundary of the set 𝒮, 𝑉̇ = 0 at
origin and boundary of set 𝒮 , 𝑉̇ is positive outside the set 𝒮 , then the system is
asymptotically stable, but not globally asymptotically stable. The set 𝒮 is
called region of attraction.
(iv) 𝑉̇ > 0 everywhere, then the system is unstable.

5. Consider the system

𝑥̇ 1 = 𝑥1 (𝑥16 + 𝑥22 − 1) + 𝑥2
𝑥̇ 2 = −𝑥1 + 𝑥2 (𝑥16 + 𝑥22 − 1)
 Consider the Lyapunov candidate function as 𝑉(𝑥) = 𝑥12 + 𝑥22 , which of the following
option(s) is/are correct. (MULTIPLE SELECT QUESTION)
A) The origin is asymptotically stable, but not globally asymptotically stable.
B) The origin is globally asymptotically stable.
C) The origin is stable in the sense of Lyapunov, but not asymptotically stable.
D) The origin is unstable.
E) None of these.

Solution:

𝑉̇ (𝑥) = (𝑥12 + 𝑥22 )(𝑥16 + 𝑥22 − 1)

For 𝑥16 + 𝑥22 < 1 , 𝑉̇ (𝑥) < 0

Here the set 𝒮 = {𝑥1 , 𝑥2 |𝑥16 + 𝑥22 ≤ 1} , is a closed set containing the origin. The
boundary of the set does not contain origin. 𝑉̇ (𝑥) is positive outside the set, zero at origin and on
the boundary of the set. 𝑉̇ (𝑥) is negative in the interior of the set except at origin.

So the system is asymptotically stable, but not globally asymptotically stable.

6. Consider the following dynamical system :

𝑥̇ 1 = −𝑥1 + (0.5)𝑥2
𝑥̇ 2 = −(0.5)𝑥2 + 3𝑠𝑖𝑛 (𝑥1 )
1 1
By taking the Lyapunov function as 𝑉 = 2 𝑥12 + 2 𝑥22 , what conclusion(s) can be drawn about
the system by taking time derivative of 𝑉 ?
A) The origin is asymptotically stable, no further analysis is required.
B) The origin is globally asymptotically stable, no further analysis is required.
C) The origin is stable in the sense of Lyapunov, but asymptotic stability can not be
concluded using the given Lyapunov function, further analysis is required.
D) The origin is unstable

Solution :
𝑥1
𝑉̇ == −𝑥12 − (0.5)𝑥22 + 𝑥2 ( + 3 sin(𝑥1 ))
2
2
𝑥22 𝑥1 𝑥2 𝑥1 + 3 sin(𝑥1 )
= −𝑥1 − + ( )
2 2 𝑥1
𝑥1 +3 sin(𝑥1 )
Define 𝐿 ≝ and 𝐿2 ≝ 𝑥1 + 3 sin(𝑥1 )
𝑥1
𝜕𝐿
To find extremum points of 𝐿2 , set 𝜕𝑥2 = 0
1
1
=> 1 − 3 cos(𝑥1 ) = 0 => 𝑥1 = cos−1 ( )
3
1
So at the extremum points, values of 𝐿2 will be 2𝑛𝜋 ± (3) ± 2√2, for 𝑛 =
0,1,2,3, … …
We can see that all the extreme points for positive 𝑥1 are positive. So we can
𝑥1 +3 sin(𝑥1 )
conclude that 𝑥1 + 3 sin(𝑥1 )is positive for positive 𝑥1 , which implies is
𝑥1
𝑥1 +3 sin(𝑥1 )
positive for positive 𝑥1 . And as 𝐿 = is an even function, we can
𝑥1
 conclude that 𝐿 is always positive. Also the maximum value of 𝐿 can be found out
to be 4, which occurs, when 𝑥1 is very close to 0.

So now we have 0 < 𝐿 ≤ 4

Also from expression of 𝐿, we can see that value of 𝐿 is approximately 1 for larger
values of 𝑥1 . So there must be a positive number 𝑎 , such that for |𝑥1 | < 𝑎, −𝑥12 −
𝐿
(0.5)𝑥22 + ( ) 𝑥1 𝑥2 is negative definite. Because negative definiteness of “−𝑥12 −
2
𝐿
(0.5)𝑥22 + ( ) 𝑥1 𝑥2 ” requires 𝐿 < 2√2, which can be possible for some positive
2
number 𝑎, as 𝐿 goes to 1 when 𝑥1 goes to infinity.

For very very small |𝑥1 |, 𝑠𝑖𝑛 (𝑥1 ) can be approximated to be 𝑥1

So the 𝑉̇ can be approximated to be
𝑥2
̇𝑉 ≈ −𝑥12 − 2 + 2𝑥1 𝑥2
2
So the worst case 𝑉̇ is −𝑥12 − 𝑥22 /2 + 2𝑥1 𝑥2
Now for a given value of 𝑥1 , let's find out the for what range of 𝑥2 ,for which " −
𝑥12 − 𝑥22 /2 + 2𝑥1 𝑥2 " will be a negative definite function.
−𝑥12 − 𝑥22 /2 + 2𝑥1 𝑥2 < 0
=> 𝑥22 − 4𝑥1 𝑥2 + 2𝑥12 > 0
This can be satisfied if 𝑥2 ∈ (−∞, (2 − √2)𝑥1 )⋃((2 + √2)𝑥1 , ∞) for 𝑥1 > 0 and
𝑥2 ∈ (−∞, (2 + √2)𝑥1 )⋃((2 − √2)𝑥1 , ∞) for 𝑥1 < 0.
So we can conclude that there is a small region around origin , for which the time
derivative of given Lyapunov function is not negative definite , but it is surely
negative definite outside that region. So using the given Lyapunov function we can
only conclude that the origin is stable in the sense of Lyapunov, but asymptotic
stability can not be concluded. For that further analysis is needed(A different
Lyapunov function should be chosen).

7. Consider the dynamical system :

1 + 2𝑥12
𝑥̇ 1 = −𝑥1 + 𝑥2 ( )
1 + 𝑥12
𝑥1 𝑥2
𝑥̇ 2 = − − 2
2√1 + 2𝑥12 1 + 𝑥2
1 𝑥12
Take the Lyapunov function as 𝑉 = + 𝑥22 . The dynamical system is :
2
√1+2𝑥12

A) Not stable in the sense of Lyapunov.

B) Stable in the sense of Lyapunov, but not asymptotically stable.
C) Asymptotically stable, but not globally asymptotically stable.
D) Globally asymptotically stable.
 Solution :
1 𝑥12
𝑉= + 𝑥22
2 √1 + 2𝑥12
𝜕𝑉 𝑥1 1 + 𝑥12
= ( )
𝜕𝑥1 √1 + 2𝑥12 1 + 2𝑥12

𝜕𝑉 1 + 2𝑥12 𝜕𝑉 𝑥1 𝑥2
=> 𝑉̇ = (−𝑥1 + 𝑥2 ( 2 )) + (− − 2)
𝜕𝑥1 1 + 𝑥1 𝜕𝑥2 2√1 + 2𝑥12 1 + 𝑥2
−𝑥12 1 + 𝑥12 𝑥1 𝑥2 𝑥1 𝑥2 2𝑥22
= ( 2) + − − 2
√1 + 2𝑥12 1 + 2𝑥1 √1 + 2𝑥12 √1 + 2𝑥12 1 + 𝑥2
−𝑥12 1 + 𝑥12 2𝑥22
= ( 2) −
√1 + 2𝑥12 1 + 2𝑥1 1 + 𝑥22

As 𝑉̇ is negative definite, so the system is globally asymptotically stable.

8. Consider the system :

𝑥̇ = 5𝑡/(1+𝑡) 𝑥 sin(𝑡) − 5𝑥
Choose the correct statement(s). (MULTIPLE SELECT QUESTION)
A) The origin is asymptotically stable.
B) The origin is globally asymptotically stable.
C) The origin is stable in sense of Lyapunov, but not asymptotically stable.
D) The origin is unstable.

Solution :

Take 𝑉(𝑥) = 0.5𝑥 2

=> 𝑉̇ (𝑥) = (5𝑡/(1+𝑡) sin(𝑡) − 5)𝑥 2

Maximum value of 𝑡/(1 + 𝑡) and sin(𝑡) are 1 . So maximum value of 5𝑡/(1+𝑡) sin(𝑡) − 5 is
zero, and it happens at 𝑡 = ∞ . Theoretically 5𝑡/(1+𝑡) never becomes 5 . It tends to 5 as time
tends to infinity. Also a sin (𝑡) term is multiplied to it. So (5𝑡/(1+𝑡) sin(𝑡) − 5) never
becomes zero. So 𝑉̇ (𝑥) is negative definite. So it is asymptotically stable. As 𝑉(𝑥) is radially
unbounded, so the system is globally asymptotically stable as well.
9. Consider the dynamical system :
𝑥̇ 1 = 𝑥2 + 5𝑥23
𝑥̇ 2 = −5𝑥1 − 𝑥23
With 𝑥1 , 𝑥2 ∈ 𝑅 .What is the number of equilibrium points of the dynamical system ?
A) 1
B) 2
C) 3
D) 4
E) 5
F) 6
G) ∞
 Solution :
𝑥2 + 5𝑥23 = 0 => 𝑥2 = 0 OR 𝑥22 = −1/5
So 𝑥2 = 0 is the only real solution. At 𝑥2 = 0 , to have 𝑥̇ 2 = −5𝑥1 − 𝑥23 = 0, we need 𝑥1 =
0.
So (0,0) is the only equilibrium point.

10. By linearizing about all the equilibrium points, what can be concluded regarding the stability
of the system about the equilibrium points?
A) All equilibrium points are stable.
B) All equilibrium points are unstable.
C) Some are equilibrium points stable, some equilibrium points are unstable.
D) No conclusion.

Solution :

0 1 + 15𝑥22
The jacobian matrix is : 𝐽 = [ ]
−5 −3𝑥22
0 1
At the equilibrium point : 𝐽 = [ ]
−5 0
The characteristics equation of 𝐽 is 𝜆2 + 5 = 0

=> 𝜆 = ±𝑖√5
As both the eigen values are purely imaginary, so we can not conclude anything about
stability using linearization method.