Systems of linear ordinary differential

equations: solved exercises

Esercizio 1.
Let (x(t), y(t)) be the solution of the Cauchy problem

 ′
x = 3x
y ′ = −x2


x(1) = 2, y(1) = 0.

Set v(t) = ex(t)y(t) and compute v ′ (1).

Exercise 2. Consider the system of differential equations

{
x′ = 3x − y
y ′ = x + y.

Set v(t) = x(2t) and w(t) = y(t), and write the system of differential equation satisfied by v and
w.

Exercise 3. Solve the system of differential equations, through its conversion into a second order
differential equation. {
x′ = y
y ′ = −3x − 4y.

∫
Exercise 4. Compute γ F · dP , where F is the vector field F (x, y) = (y − x2 , 2x − 1) and γ(t),
t ∈ [0, 1], is a solution curve of the system of differential equations
{
x′ = 2x − 1
y ′ = x2 − y.

⃝
c 2018 Politecnico di Torino
2 Systems of linear ordinary differential equations: solved exercises

Exercise 5. Consider the systems of ordinary differential equations

{ √ { √
x′ = x |t| x′ = |y − t|
(a) (b)
y ′ = |y|y y ′ = x.

In both cases, find out for which points (t0 , x0 , y0 ) ∈ R3 the hypotheses of the Cauchy Theorem
of local existence and uniqueness are verified.

Exercise 6. Consider the following system of linear ordinary differential equations:

{
ẋ = 3x − 2y
ẏ = 2x − 2y.

a) Determine the general solution of the system.

b) Find the non-trivial solutions, if any, bounded in the future, that is bounded on [0, +∞).

c) Determine the solution of the Cauchy problem, with initial conditions x(0) = 1, y(0) = 2.

Exercise 7. Consider the following system of linear ordinary differential equations:

{
ẋ = 2x + y
ẏ = −x + 2y.

a) Determine the general solution of the system.

b) Find the non-trivial solutions, if any, bounded in R.

c) Find the non-trivial solutions, if any, bounded in the past, that is bounded on (−∞, 0].

d) Determine the solution of the Cauchy problem, with initial conditions x(0) = 3, y(0) = 2.

⃝
c 2018 Politecnico di Torino
Systems of linear ordinary differential equations: solved exercises 3

Exercise 8. Consider the following system of linear ordinary differential equations:

{
ẋ = 3x + y
ẏ = −x + y.

a) Determine the general solution of the system.

b) Find the non-trivial solutions, if any, bounded in the future, that is bounded on [0, +∞).

c) Find the non-trivial solutions, if any, bounded in the past, that is bounded on (−∞, 0].

d) Determine the solution of the Cauchy problem, with initial conditions x(0) = −1, y(0) = 3.

Exercise 9. Consider the following system of linear ordinary differential equations:

{
ẋ = x + 2y
ẏ = −2x − 4y.

a) Determine the general solution of the system.

b) Find the non-trivial solutions, if any, bounded in R.

c) Find the non-trivial solutions, if any, bounded in the future, that is bounded on [0, +∞).

d) Find the non-trivial solutions, if any, bounded in the past, that is bounded on (−∞, 0].

e) Determine the solution of the Cauchy problem, with initial conditions x(0) = 0, y(0) = 3.

f ) Determine the solution of the Cauchy problem, with initial conditions x(0) = −4, y(0) = 2.

Exercise 10. Consider the following system of linear ordinary differential equations:
{
ẋ = x + 2y
ẏ = −2x − y.

a) Determine the general solution of the system.

b) Find the non-trivial solutions, if any, bounded in R.

c) Determine the solution of the Cauchy problem, with initial conditions x(0) = 2, y(0) = 0.

⃝
c 2018 Politecnico di Torino
4 Systems of linear ordinary differential equations: solved exercises

Exercise 11. Determine the general solution of the following systems of non-homogeneous linear
differential equations:
{ {
ẋ = −x + 4y + e3t ẋ = −x + 4y + 2 + e3t
(a) (b)
ẏ = −x + 3y ẏ = −x + 3y − 1

Exercise 12. Determine the solution of the Cauchy problem



ẋ = −x + 4y + cos t
ẏ = −x + 3y


x(0) = 0, y(0) = 0.

Exercise 13. Determine the solution of the following Cauchy problems:

 
 −t  t
ẋ = x + 2y + e ẋ = x + 2y + e
(a) ẏ = 2x + y (b) ẏ = 2x + y + 1

 

x(0) = 0, y(0) = 0 x(0) = 0, y(0) = 0

Exercise 14. Study the stability of the zero solution of the systems given in Exercises 6-10.
Classify the origin as a critical point.

Exercise 15. Study the stability of the system as α varies in R. In each case, classify the origin
as a critical point.
{ { {
ẋ = −3x + α2 y ẋ = 2y ẋ = αx − y
(a) (b) (c)
ẏ = x − 3y ẏ = −2x + αy ẏ = x + y

⃝
c 2018 Politecnico di Torino
Systems of linear ordinary differential equations: solved exercises 5

Exercise 16. Consider the system of linear differential equations,

{
ẋ = x + α2 y
ẏ = α1 x + αy.

a) Study the stability of the system as α varies in R \ {0}.

b) Find, if any, the values of α such that all the solutions are bounded in R.

c) Find, if any, the values of α such that α there exist solutions bounded in [0, +∞).

Exercise 17. Consider the system of linear differential equations,

{
ẋ = (α + 1)x + (α2 − 1)y
ẏ = x + αy.

a) Study the stability of the system as α varies in R. In each case, classify the origin as a critical
point.

b) Find, if any, the values of α such that all the solutions are bounded in R.

c) Find, if any, the values of α such that α there exist solutions bounded in [0, +∞).

d) Find, if any, the values of α such that α there exist solutions bounded in (−∞, 0].

Exercise 18. Consider the system of linear differential equations,

{
ẋ = (α + 2)x + y
ẏ = −x + αy.

a) Study the stability of the system as α varies in R. In each case, classify the origin as a critical
point.

b) Study the boundedness in the past and in the future of the solutions, as α varies in R.

c) For α = 1, determine the general solution of the system.

⃝
c 2018 Politecnico di Torino
6 Systems of linear ordinary differential equations: solved exercises

Exercise 19. Consider the system of linear differential equations,

{
ẋ = x
ẏ = −y − 2x.

a) Compute its generale solution.

b) Study the stability of the origin.

c) Study the boundedness in the future and in the past of the solutions.

d) Find the values α ∈ R such that the solutions (x(t), y(t)) verify
( ) ( )
αt x(t) 0
lim e = .
t→+∞ y(t) 0

e) Determine the solution satisfying the initial condition x(0) = 2, y(0) = −2.

⃝
c 2018 Politecnico di Torino
Systems of linear ordinary differential equations: solved exercises 7

SOLUTIONS

Exercise 1. We derive v(t) = ex(t)y(t) and we use the equations of the system to get
[ ]
v ′ (t) = ex(t)y(t) x′ (t)y(t) + x(t)y ′ (t)
[ ]
= ex(t)y(t) 3x(t)y(t) − [x(t)]3 .

Since x(1) = 2 and y(1) = 0, if we substitute to t the value 1, we obtain

[ ]
v ′ (1) = e2·0 3 · 2 · 0 − 23 = −8.

Exercise 2. We derive with respect to t the functions v(t) = x(2t) and w(t) = y(t), then we use
the equations of the system, to get:

v ′ (t) = 2x′ (2t) = 2 [3x(2t) − y(2t)] =

= 6v(t) − 2w(2t)

w′ (t) = y ′ (t) = x(t) + y(t) =

= v(t/2) + w(t).

Thus, we find the system of differential equations

{
v ′ (t) = 6v(t) − 2w(2t)
w′ (t) = v(t/2) + w(t).

Note that this system isn’t a system of ordinary differential equations.

Exercise 3.
If we derive x′ = y, we get x′′ = y ′ . If we substitute in the second equation, we get

x′′ = y ′ = −3x − 4x′ =⇒ x′′ + 4x′ + 3x = 0.

This is a second order, homogeneous linear differential equation with constant coefficients. To solve
it, we consider the associated characteristic equation:

z 2 + 4z + 3 = 0,

⃝
c 2018 Politecnico di Torino
8 Systems of linear ordinary differential equations: solved exercises

whose solutions are z1 = −1 and z2 = −3. The general solution of the second order equation is

x(t) = c1 e−t + c2 e−3t , c1 , c2 ∈ R.

If we derive, we get x′ (t) = y(t) = −c1 e−t − 3c2 e−3t . Thus, the general integral of the system is,
{
x(t) = c1 e−t + c2 e−3t
c1 , c2 ∈ R.
y(t) = −c1 e−t − 3c2 e−3t ,

Exercise 4.
Remark that γ ′ (t) = (2x(t) − 1, [x(t)]2 − y(t)) and F (x(t), y(t)) = (y(t) − [x(t)]2 , 2x(t) − 1), so
that F (γ(t)) · γ ′ (t) = 0 for all t ∈ [0, 1]. Thus, the two vector are orthogonal, and
∫ ∫ 1
F · dP = F (γ(t)) · γ ′ (t) dt = 0.
γ 0

Exercise 5.
Let {
x′ = f1 (t, x, y)
y ′ = f2 (t, x, y)

be a system of differential equations, with initial condition x(t0 ) = x0 , y(t0 ) = y0 . Cauchy Theorem
guarantees the existence and uniqueness of local solutions under the hypotheses of continuity in a
∂fi ∂fi
neighbourhood of (t0 , x0 , y0 ) of the functions fi and of the partial derivatives , , for i = 1, 2.
∂x ∂y
√
5a) f1 (t, x, y) = x |t| and f2 (t, x, y) = y|y| are continuous in R3 . Moreover,

∂f1 √
(t, x, y) = |t|
∂x
∂f1 ∂f2
(t, x, y) = (t, x, y) = 0
∂y ∂x
∂f2
(t, x, y) = 2|y|
∂y

are continuous in R3 . Therefore the hypotheses of Cauchy Theorem are verified for all
(t0 , x0 , y0 ) ∈ R3 .

⃝
c 2018 Politecnico di Torino
Systems of linear ordinary differential equations: solved exercises 9

√
5b) f1 (t, x, y) = |y − t| and f2 (t, x, y) = x are continuous in R3 . Moreover,
∂f1
(t, x, y) = 0
∂x
∂f1 sgn(y − t)
(t, x, y) = √
∂y 2 |y − t|
∂f2
(t, x, y) = 1
∂x
∂f2
(t, x, y) = 0.
∂y

The derivative of f1 with respect to y is not defined at t0 = y0 , while it is continuous elsewhere.

The other derivatives are everywhere continuous.

Thus, the hypotheses of Cauchy Theorem aren’t verified on the plane y = t and are verified
elsewhere.

Exercise 6. To compute the general solution of a system of homogeneous linear differential

equations, with constant coefficients, Ẋ = AX, first of all we need to compute the eigenvalues of
the matrix A, that is the solutions of the characteristic equation det(A − λI) = 0. We recall that,
if A is a 2 × 2 matrix, we have det(A − λI) = λ2 − tr Aλ + det A = 0

6a) The matrix form of the system is

( ) ( ) ( )( )
ẋ x 3 −2 x
=A = .
ẏ y 2 −2 y
Since tr A = 1 and det A = −2, the characteristic equation is

λ2 − λ − 2 = (λ − 2)(λ + 1) = 0.

Hence the eigenvalues of A are λ1 = 2 and λ2 = −1. If we denote by ui an eigenvector

associated to the eigenvalue λi (i=1,2), the general solution is
( )
x(t)
= c1 e2t u1 + c2 e−t u2 , c1 , c2 ∈ R.
y(t)
Now we find the eigenvectors: for each eigenvalue, we solve the equation (A − λi I)u = 0.
( )
a
• λ1 = 2 If we set u = , we obtain the linear system
b
( )( ) ( )
1 −2 a 0
= ,
2 −4 b 0

⃝
c 2018 Politecnico di Torino
10 Systems of linear ordinary differential equations: solved exercises

which corresponds to the equation a − 2b = 0. Hence an eigenvector associated to the

eigenvalue λ1 = 2 is ( )
2
u1 = .
1

• λ2 = −1 In this case the associated linear system is

( )( ) ( )
4 −2 a 0
= ,
2 −1 b 0

which corresponds to the equation 2a − b = 0. Hence an eigenvector associated to the

eigenvalue λ2 = −1 is ( )
1
u2 = .
2

The general solution of the system is

( ) ( ) ( )
x(t) 2 −t 1
= c1 e 2t
+ c2 e , c 1 , c2 ∈ R,
y(t) 1 2

which can be written also as

{
x(t) = 2c1 e2t + c2 e−t
c1 , c2 ∈ R.
y(t) = c1 e2t + 2c2 e−t

6b) We recall that a solution is bounded in an interval I if there exists M > 0 such that
√
∥(x(t), y(t))∥ = x2 (t) + y 2 (t) ≤ M, ∀t ∈ I.

Since, for all t ∈ I,

∥(x(t), y(t))∥ ≤ |x(t)| + |y(t)|,

if x(t) and y(t) are bounded in I, (x(t), y(t)) is bounded as well.

We recall that, if a function x(t) is continuous on a half-line [0, +∞) and it is such that
lim x(t) ∈ R, then x(t) is bounded in [0, +∞). To prove this statement we use Weierstrass
t→+∞
Theorem and the definition of limit.

Analogously we can prove that, if x(t) is continuous in (−∞, 0] and lim x(t) ∈ R, then x(t)
t→−∞
is bounded in (−∞, 0].

If x(t) is continuous in R and the limits of x(t) as t → +∞ and as t → −∞ are finite, then
x(t) is bounded in R.

Anytime the study of the boundedness of the solutions is required, we’ll use these results.

⃝
c 2018 Politecnico di Torino
Systems of linear ordinary differential equations: solved exercises 11

We consider now the case of Exercise 6. We remark that

{
( 2t
)
+∞ if c1 ̸= 0
−t
lim x(t) = lim 2c1 e + c2 e =
t→+∞ t→+∞ 0 if c1 = 0
{
( ) +∞ if c1 ̸= 0
lim y(t) = lim c1 e2t + 2c2 e−t =
t→+∞ t→+∞ 0 if c1 = 0.

Hence x(t) and y(t) are bounded in [0, +∞) if and only if c1 = 0. Therefore, there exist ∞1
solutions (the ones for c1 = 0) bounded in [0, +∞).

6c) The solution must satisfy the initial condition x(0) = 1, y(0) = 2; if we substitute to t the
value 0 in the general solution, we get
{
x(0) = 2c1 + c2 = 1
y(0) = c1 + 2c2 = 2,

whose solutions are c1 = 0, c2 = 1. Hence, the solution of the Cauchy problem is

{
x(t) = e−t
y(t) = 2e−t .

Exercise 7.

7a) The matrix form of the system is

( ) ( ) ( )( )
ẋ x 2 1 x
=A = .
ẏ y −1 2 y

From the characteristic equation

det(A − λI) = (2 − λ)2 + 1 = 0

we get (2 − λ)2 = −1, therefore the eigenvalues of A are λ1,2 = 2 ± i.

Since they are complex numbers, the associated eigenvectors are two complex conjugate
vectors u ± iv, and the general solution is
( )
x(t)
= c1 e2t [cos tu − sin tv] + c2 e2t [sin tu + cos tv] , c 1 , c2 ∈ R.
y(t)

To compute(w =)u + iv, we have to solve the linear system (A − (2 + i)I)w = 0, that is, if
a
we set w = , with a, b ∈ C:
b

⃝
c 2018 Politecnico di Torino
12 Systems of linear ordinary differential equations: solved exercises

( )( ) ( )
−i 1 a 0
= ,
−1 −i b 0
which corresponds to the equation −ia + b = 0. If we choose for instance a = 1 we find the
complex eigenvalue ( ) ( ) ( )
1 1 0
w= = +i = u + iv.
i 0 1

The general solution of the system is

( ) [ ( ) ( )] [ ( ) ( )]
x(t) 1 0 1 0
2t
= c1 e cos t − sin t 2t
+ c2 e sin t + cos t
y(t) 0 1 0 1

with c1 , c2 ∈ R, which can be written as,

{
x(t) = e2t [c1 cos t + c2 sin t]
c1 , c2 ∈ R.
y(t) = e2t [−c1 sin t + c2 cos t]

7b) Since
lim |x(t)| = lim e2t |c1 cos t + c2 sin t| = +∞,
t→+∞ t→+∞

x(t) is unbounded in R, hence there are no non-trivial solutions, bounded in R.

7c) We remark that, for all t ∈ (−∞, 0]

|x(t)| = |e2t [c1 cos t + c2 sin t] | ≤ e2t (|c1 | + |c2 |) ≤ |c1 | + |c2 |
|y(t)| = |e2t [−c1 sin t + c2 cos t] | ≤ e2t (|c1 | + |c2 |) ≤ |c1 | + |c2 |

Hence all the solutions are bounded in the past.

7d) The solution must satisfy the initial condition x(0) = 3, y(0) = 2; if we substitute to t the
value 0 in the general solution, we get
{
x(0) = c1 = 3
y(0) = c2 = 2.

Hence the solution of the Cauchy problem is

{
x(t) = e2t [3 cos t + 2 sin t]
y(t) = e2t [−3 sin t + 2 cos t] .

⃝
c 2018 Politecnico di Torino
Systems of linear ordinary differential equations: solved exercises 13

Exercise 8.

8a) The matrix form of the system is

( ) ( ) ( )( )
ẋ x 3 1 x
=A = .
ẏ y −1 1 y

Since tr A = 4 and det A = 4, the characteristic equation is

λ2 − 4λ + 4 = (λ − 2)2 = 0.

Hence A has a unique eigenvalue, λ = 2, with

( algebraic
) multiplicity 2. We have to solve the
a
linear system (A − 2I)u = 0. If we set u = , we get the linear system
b
( )( ) ( )
1 1 a 0
= ,
−1 −1 b 0

which corresponds to the equation a+b = 0. Hence the eigenspace associated to the eigenvalue
2 has dimension 1, that is λ = 2 has geometric molteplicity 1. An eigenvector associated to
the eigenvalue λ = 2 is ( )
1
u= .
−1
To find a second linearly independent solution, we have to find a(generalized
) eigenvector, that
a
is a solution of the linear system (A − 2I)v = u. If we set v = , we get
b
( )( ) ( )
1 1 a 1
= ,
−1 −1 b −1

which corresponds to the equation a + b = 1. If we choose a = 0, we get the generalized

eigenvector ( )
0
v= .
1

The general solution of the system is

( ) ( ) [ ( ) ( )]
x(t) 1 1 0
= c1 e 2t 2t
+ c2 e t + , c1 , c2 ∈ R,
y(t) −1 −1 1

which can be written as,

{
x(t) = c1 e2t + c2 te2t = e2t (c1 + c2 t)
c1 , c2 ∈ R.
y(t) = −c1 e2t − c2 te2t + c2 e2t = e2t (−c1 + c2 − c2 t)

⃝
c 2018 Politecnico di Torino
14 Systems of linear ordinary differential equations: solved exercises

8b) In this case

 
lim |x(t)| = lim c1 e2t + c2 te2t  = lim e2t |c1 + c2 t| = +∞
t→+∞ t→+∞ t→+∞
 
lim |y(t)| = lim −c1 e − c2 te + c2 e2t  = lim e2t | − c1 + c2 − c2 t| = +∞.
2t 2t
t→+∞ t→+∞ t→+∞

Hence there are no non-trivial solutions, bounded in [0, +∞). We have

lim |x(t)| = lim e2t |c1 + c2 t| = 0

t→−∞ t→−∞
lim |y(t)| = lim e2t | − c1 + c2 − c2 t| = 0.
t→−∞ t→−∞

Hence all the solutions are bounded in the past.

8d) The solution must satisfy the initial condition x(0) = −1, y(0) = 3; if we substitute to t the
value 0 in the general solution, we get
{
x(0) = c1 = −1
y(0) = −c1 + c2 = 3,

whose solutions are c1 = −1, c2 = 2. Hence the solution of the Cauchy problem is
{
x(t) = −e2t + 2te2t
y(t) = e2t − 2te2t + 2e2t

Exercise 9.

9a) The matrix form of the system is

( ) ( ) ( )( )
ẋ x 1 2 x
=A = .
ẏ y −2 −4 y

Since tr A = −3 and det A = 0, the characteristic equation is

λ2 + 3λ = λ(λ + 3) = 0.

therefore the eigenvalues are λ1 = 0 and λ2 = −3. Now we compute the eigenvectors associ-
ated to these real eigenvalues.
( )
a
• λ1 = 0 If we set u = , we get the linear system
b
( )( ) ( )
1 2 a 0
= ,
−2 −4 b 0

⃝
c 2018 Politecnico di Torino
Systems of linear ordinary differential equations: solved exercises 15

which corresponds to the equation a + 2b = 0. Hence an eigenvector associated to the

eigenvalue λ1 = 0 is ( )
−2
u1 = .
1

• λ2 = −3 The associated linear system is

( )( ) ( )
4 2 a 0
= ,
−2 −1 b 0

which corresponds to the equation −2a − b = 0. Hence an eigenvector associated to the

eigenvalue λ2 = −3 is ( )
1
u2 = .
−2

The general solution of the system is

( ) ( ) ( )
x(t) −2 −3t 1
= c1 + c2 e , c1 , c2 ∈ R,
y(t) 1 −2

which can be written as,

{
x(t) = −2c1 + c2 e−3t
c1 , c2 ∈ R.
y(t) = c1 − 2c2 e−3t

9b) In this case

{
 +∞ if c2 ̸= 0
lim |x(t)| = lim −2c1 + c2 e−3t  =
t→−∞ t→−∞ |2c1 | if c2 = 0
{
  +∞ if c2 ̸= 0
lim |y(t)| = lim c1 − 2c2 e−3t  =
t→−∞ t→−∞ |c1 | if c2 = 0

while
 
lim |x(t)| = lim −2c1 + c2 e−3t  = |2c1 |
t→+∞ t→+∞
 
lim |y(t)| = lim c1 − 2c2 e−3t  = |c1 |.
t→+∞ t→+∞

Therefore there exist ∞1 solutions bounded in R, corresponding to c2 = 0. These solutions

form a line of critical points, parallel to the vector of components (−2, 1).

9c) Since the limits as t → +∞ of |x(t)| and |y(t)| are finite, all the solutions are bounded in the
future.

9d) We already remarked that the solutions bounded in the past are those such that c2 = 0.

⃝
c 2018 Politecnico di Torino
16 Systems of linear ordinary differential equations: solved exercises

9e) The solution must satisfy the initial condition x(0) = 0, y(0) = 3; if we substitute to t the
value 0 in the general solution, we get
{
x(0) = −2c1 + c2 = 0
y(0) = c1 − 2c2 = 3,
whose solution is c1 = −1, c2 = −2. Hence the solution of the Cauchy problem is
{
x(t) = 2 − 2e−3t
y(t) = −1 + 4e−3t

9f ) The solution must satisfy the initial condition x(0) = −4, y(0) = 2; observe that (−4, 2) =
2u1 t = 2(−2, 1), which corresponds to the constant solution x(t) = −4, y(t) = 2, for all t ∈ R.

Exercise 10.

10a) The matrix form of the system is

( ) ( ) ( )( )
ẋ x 1 2 x
=A = .
ẏ y −2 −1 y
The characteristic equation is
λ2 + 3 = 0,
√
hence A has the two complex conjugate eigenvalues λ1,2 = ±i 3.

As we already remarked in Exercise 7, the associated eigenvectors are complex conjugate

√
vectors u ± iv. To compute
( ) w = u + iv, we have to solve the linear system (A − i 3I)w = 0,
a
that is, if we set w = , with a, b ∈ C:
b
( √ )( ) ( )
1−i 3 2 √ a 0
= ,
−2 −1 − i 3 b 0
√
which corresponds to the equation (1 − i 3)a + 2b = 0. If we choose a = 2 we find the
complex eigenvector
( ) ( ) ( )
2 √ 2 √0
w= = +i = u + iv.
−1 + i 3 −1 3
The general solution of the system is
( ) [ ( ) ( )] [ ( ) ( )]
x(t) √ 2 √ 0 √ 2 √ 0
= c1 cos 3t − sin 3t √ +c2 sin 3t + cos 3t √
y(t) −1 3 −1 3
with c1 , c2 ∈ R, which can be written as,
{ √ √
x(t) = 2c1 cos 3t + 2c2 sin 3t
√ √ √ √ c1 , c2 ∈ R.
y(t) = (−c1 + 3c2 ) cos 3t + (−c1 3 − c2 ) sin 3t,

⃝
c 2018 Politecnico di Torino
Systems of linear ordinary differential equations: solved exercises 17

10b) In this case, for all t ∈ R, we have

|x(t)| ≤ 2|c1 | + 2|c2 |

√ √
|y(t)| = | − c1 + 3c2 | + | − c1 3 − c2 |.

Therefore all the solutions are bounded in R. Moreover, all the non-trivial solutions are
periodic, with minimal period 2π/3.

10c) The solution must satisfy the initial condition x(0) = 2, y(0) = 0; if we substitute in the
general solution, we get {
x(0) = 2c1 = 2
√
y(0) = −c1 + 3c2 = 0,
√
1 3
whose solution is c1 = 1, c2 = √ = . Hence the solution of the Cauchy problem is
3 3
 √
 √ 2 3 √
x(t) = 2 cos 3t + sin 3t
√ 3
y(t) = − 4 3 sin √3t.

3

Exercise 11. The two systems are systems of non-homogeneous, linear differential equations. We
recall that the general solution of these systems is obtained by summing to the general solution
of the associated homogeneous system a particular solution of the non-homogeneous system. To
compute this one, we can use the results seen in class.
In both systems, the associated homogeneous system is
( ) ( ) ( )( )
ẋ x −1 4 x
=A = .
ẏ y −1 3 y

The characteristic equation is λ2 − 2λ + 1 = (λ − 1)2 = 0, which has a unique eigenvalue, λ = 1,

with algebraic multiplicity 2 (see exercise 8). ( )
a
We solve the equation (A − I)u = 0. If we set u = , we get the linear system
b
( )( ) ( )
−2 4 a 0
= ,
−1 2 b 0
which corresponds to the equation −a + 2b = 0. Hence the eigenspace associated to the eigenvalue
1 has dimension 1, that is λ = 1 has geometric multiplicity 1. An eigenvector associated to the
eigenvalue λ = 1 is ( )
2
u= .
1

⃝
c 2018 Politecnico di Torino
18 Systems of linear ordinary differential equations: solved exercises

To determine a second, linearly independent solution, we look

( for)a generalized eigenvector, that is
a
a solution of the linear system (A − I)v = u. If we set v = , we get
b
( )( ) ( )
−2 4 a 2
= ,
−1 2 b 1

which corresponds to the equation −a + 2b = 1. If we choose b = 0, we get the generalized

eigenvector ( )
−1
v= .
0
The general solution of the system is
( ) ( ) [ ( ) ( )]
x(t) 2 2 −1
= c1 e t t
+ c2 e t + , c 1 , c2 ∈ R,
y(t) 1 1 0

which can be written as,

{
x(t) = 2c1 et + c2 et (2t − 1) = et (2c1 − c2 + 2t)
c1 , c2 ∈ R.
y(t) = c1 et + c2 tet = et (c1 + c2 t)

11a) Now we look for a solution of the non-homogeneous system:

( ) ( ) ( )( ) ( )
ẋ x −1 4 x 3t 1
=A = +e .
ẏ y −1 3 y 0

The forcing term is such as eµt P(t), where µ = 3 and P(t) is a vector of polynomials of degree
0. Since 3 isn’t an eigenvalue of A, there exists a solution such as
( ) ( )
x(t) 3t 3t a
= e Q(t) = e .
y(t) b

If we derive and substitute into the system, the following system has to be satisfied:
{
3ae3t = −ae3t + 4be3t + e3t = e3t (−a + 4b + 1)
3be3t = −ae3t + 3be3t = e3t (−a + 3b).

1
If we divide by e3t we get a linear system, whose solutions are a = 0 and b = − .
4
Therefore the general solution of the non-homogeneous system is:

x(t) = 2c1 et + c2 et (2t − 1)
1 c1 , c2 ∈ R.
y(t) = c1 et + c2 tet − e3t .
4

⃝
c 2018 Politecnico di Torino
Systems of linear ordinary differential equations: solved exercises 19

11b) In this case the forcing term is

( ) ( )
1 2
e3t + ,
0 −1

which is the sum of the forcing term of exercise (11a) with a constant forcing term, that can
be interpreted as a vector of polynomials of degree 0. To find a solution of this system, we
use the superposition principle; therefore, we need to find a solution of the system
( ) ( )( ) ( )
ẋ −1 4 x 2
= + ,
ẏ −1 3 y −1

that we’ll sum to the previously found solution of the first system.

Since ( ) ( )
2 2
= e0t ,
−1 −1
and 0 isn’t an eigenvalue of A, there exists a solution such as
( ) ( ) ( )
x(t) a a
= e0t = .
y(t) b b

If we derive and we substitute in the differential system, we see that a, b must satisfy the
following linear system: {
0 = −a + 4b + 2
0 = −a + 3b − 1
whose solutions are a = −10 and b = −3. Hence we obtain a solution of the system by
summing up this vector with the solution found before, that is:
( ) ( ) ( )
x(t) 3t 0 −10
=e + .
y(t) − 41 −3

The general solution of the differential system is:


x(t) = 2c1 et + c2 et (2t − 1) − 10
1 c1 , c2 ∈ R.
y(t) = c1 et + c2 tet − e3t − 3.
4

Exercise 12. The homogeneous differential system associated to the complete one is the same as
that of the previous exercise. The forcing term is
( ) ( )
cos t cos t
= e0t P(t) .
0 0

⃝
c 2018 Politecnico di Torino
20 Systems of linear ordinary differential equations: solved exercises

Since 0 + i isn’t an eigenvalue of A, there exists a solution such as

( ) ( ) ( ) ( )
x(t) a b a cos t + b sin t
= cos t + sin t = .
y(t) c d c cos t + d sin t
If we derive and we substitute into the differential system, we get
{
−a sin t + b cos t = −(a cos t + b sin t) + 4(c cos t + d sin t) + cos t
−c sin t + d cos t = −(a cos t + b sin t) + 3(c cos t + d sin t)

which is equivalent to the system

{
−a sin t + b cos t = (−b + 4d) sin t + (−a + 4c + 1) cos t
−c sin t + d cos t = (−b − 3d) sin t + (−a + 3c) cos t.

Two trigonometric polynomial in sin t and cos t equal if and only if the coefficients of sin t and cos t
are the same; thus, we get the linear system of four equations in four unknowns
 

 −a = −b + 4d 
 a = − 21

 

 b = −a + 4c + 1 b = 3
=⇒ 2

 −c = −b + 3d 
 c = 0

 

 d = −a + 3c d = 1
2

Hence the general solution of the differential system is

{
x(t) = 2c1 et + c2 et (2t − 1) − 12 cos t + 32 sin t
c1 , c2 ∈ R.
y(t) = c1 et + c2 tet + 12 sin t.

To determine the solution satisfying the initial condition x(0) = 0, y(0) = 0, we substitute to
t the value 0 in the general solution. We get the linear system
{
2c1 − c2 − 21 = 0
c1 = 0

whose solutions are c1 = 0 and c2 = − 12 . Therefore the solution of the Cauchy problem is:
{
x(t) = − 12 et (2t − 1) − 12 cos t + 32 sin t
y(t) = − 21 tet + 12 sin t.

Exercise 13. In both exercises, the associated homogeneous differential system is:
( ) ( ) ( )( )
ẋ x 1 2 x
=A = .
ẏ y 2 1 y

The characteristic equation is λ2 − 2λ − 3 = (λ − 3)(λ + 1) = 0, which has the real solutions λ1 = 3

and λ2 = −1. We see that :

⃝
c 2018 Politecnico di Torino
Systems of linear ordinary differential equations: solved exercises 21

( )
1
• to the eigenvalue λ1 = 3 is associated an eigenvector u1 =
1
( )
1
• to the eigenvalue λ2 = −1 is associated an eigenvector u2 = .
−1
Therefore, the general solution of the homogeneous differential system is
{
x(t) = c1 e3t + c2 e−t
, c1 , c2 ∈ R
y(t) = c1 e3t − c2 e−t
13a) In this case the forcing term is such as e−t P(t), where P(t) is a vector of polynomials of
degree 0. Since -1 is an eigenvalue of A with algebraic multiplicity 1, there exists a solution
such as ( ) ( ) ( −t )
x(t) −t at + b e (at + b)
=e = .
y(t) ct + d e−t (ct + d)
If we derive and we substitute into the differential system, we obtain
{
x′ (t) = −e−t (at + b) + ae−t = (at + b)e−t + 2(ct + d)e−t + e−t
y ′ (t) = −e−t (ct + d) + ce−t = 2(at + b)e−t + (ct + d)e−t
If we divide by e−t we get
{
(2a + 2c)t + (2b + 2d − a + 1) = 0
(2a + 2c)t + (2b + 2d − c) = 0.
which determines a system with ∞1 solutions: we’ll choose one of them.

 
 a = 21
c = −a 

  c = − 12
2b + 2d = a − 1 =⇒

 
 d = 0 (for instance)
2b + 2d = −a 

b = − 1 .
4
Therefore the general solution of the non-homogeneous system is
{ ( )
x(t) = c1 e3t + c2 e−t + 12 t − 41 e−t
y(t) = c1 e3t − c2 e−t − 12 te−t .
Now we look for the solution satisfying the initial conditions x(0) = y(0) = 0. If we substitute
t = 0 in the general solution, we get the linear system
{
x(0) = c1 + c2 − 14 = 0
y(t) = c1 − c2 = 0
whose solution is c1 = c2 = 18 . Therefore, the solution of the Cauchy problem is:
 ( )

 1 3t 1 −t 1 1 −t

 x(t) = e + 8 e + t− e
 8 2 4



 1 1 1
y(t) = e3t − e−t − te−t .
8 8 2

⃝
c 2018 Politecnico di Torino
22 Systems of linear ordinary differential equations: solved exercises

13b) In this case the forcing term is such as et P(t) + e0t Q(t), where P(t) and Q(t) are vectors of
polynomials of degree 0. Since 1 and 0 aren’t eigenvalues of A, the superposition principle
guarantees the existence of a solution such as
( ) ( ) ( ) ( )
x(t) a t b a + et b
= +e =
y(t) c d c + et d

If we derive and we substitute into the differential system, we obtain

{ {
x′ (t) = bet = a + bet + 2(c + det ) + et (a + 2c) + et (2d + 1) = 0
=⇒
y(t) = det = 2(a + bet ) + c + det + 1 (2a + c + 1) + bet = 0.

Since the polynomials and the exponential functions are linearly independent, the equalities
are verified if and only if the constant functions and the coefficients of et equal 0. Thus, we
get a linear system
 

(a + 2c) = 0 
a = − 3
2

 

2d + 1 = 0 b = 0
=⇒

 2a + c + 1 = 0 
c = 13

 

b = 0 d = − 1 .
2

Therefore the general solution of the differential system is

{
x(t) = c1 e3t + c2 e−t − 32
y(t) = c1 e3t − c2 e−t + 13 − 12 et .

Now we look for the solution satisfying the initial conditions x(0) = y(0) = 0. If we substitute
to t the value 0 in the general solution we find the linear system
{ {
x(0) = c1 + c2 − 23 = 0 c1 = 125
=⇒
y(0) = c1 − c2 + 13 − 12 = 0 c2 = 14 .

Thus, the solution of the Cauchy problem is:

{
e + 14 e−t − 23
5 3t
x(t) = 12
e − 14 e−t + 31 − 12 et .
5 3t
y(t) = 12

Exercise 14. To study the stability of the origin (or equivalently, of the zero solution) of a system
of homogeneous, linear differential equations Ẋ = AX we study the sign of the real part of the
eigenvalues of A .

⃝
c 2018 Politecnico di Torino
Systems of linear ordinary differential equations: solved exercises 23

14a) In Exercise 7, the eigenvalues of A are λ1 = 2 > 0 and λ2 = −1 < 0. Hence the origin is an
unstable, non-repulsive critical point, that is, a saddle.

14b) In Exercise 8, A has the complex conjugate eigenvalues λ1,2 = 2±i. Since Re λ1,2 = 2 > 0, the
origin is unstable and repulsive (or completely unstable). Since the eigenvalues are complex,
with positive real part, the origin is an unstable focus.

14c) In Exercise 9, A has the double eigenvalue λ1,2 = 2 > 0. In this case the zero solution is
completely unstable (or unstable and repulsive). Since the eigenvalues are real, the origin is
an unstable node.

14d) In Exercise 10, the eigenvalues of A are λ1 = 0 and λ2 = −3 < 0. In this case the system
is stable; there are infinitely many stable, constant solutions. None of the critical points is
attractive.
√
14e) In Exercise 11, A has complex conjugate eigenvalues λ1,2 = ±i 3. Since Re λ1,2 = 0, the
zero solution is stable, non-attractive. The origin is a centre (all the orbits different from the
origin are simple, closed curves, containing the origin in its interior).

Exercise 15.

15a) We have to study the stability of the origin as α varies in R of the system
( ) ( ) ( )( )
ẋ x −3 α2 x
=A = .
ẏ y 1 −3 y

The solutions of the characteristic equation

(−3 − λ)2 − α2 = 0

are λ1,2 = −3 ± |α|. We remark that

• If |α| < 3, λ1,2 < 0.

• If |α| = 3, λ1 = −3 + 3 = 0 and λ2 = −3 − 3 = −6 < 0.

• If |α| > 3, λ1 = −3 + |α| > 0 and λ2 = −3 − |α| < 0.

Therefore we are able to study the stability of the zero solution and to classify the origin as
a critical point.

⃝
c 2018 Politecnico di Torino
24 Systems of linear ordinary differential equations: solved exercises

|α| < 3 λ1,2 < 0 O asymptotically stable O asymptotically stable node

|α| = 3 λ1 = 0, λ2 < 0 O stable, non-attractive line of critical points
|α| > 3 λ1 > 0, λ2 < 0 O unstable, non-repulsive O saddle

15b) We have to study the stability of the origin as α varies in R of the system
( ) ( ) ( )( )
ẋ x 0 2 x
=A = .
ẏ y −2 α y

The characteristic equation is:

λ2 − αλ + 4 = 0.

Since ∆ = α2 − 16 has different sign depending on the values of α, we separate the different
cases.

• ∆ = α2 − 16 < 0 if and only if |α| < 4. In this case we have two complex conjugate
eigenvalues, such that


> 0 if 0 < α < 4
α
Re λ1,2 = =0 if α = 0
2 

<0 if − 4 < α < 0

• ∆ = α2 − 16 = 0 if and only if |α| = 4. In this case we have a real, double eigenvalue

and {
|α| 2 > 0 if α = 4
λ= :
2 −2 < 0 if α − 4

• ∆ = α2 −16 > 0 if and only if |α| > 4. We recall that, if λ1,2 are the solutions of a second
order algebraic equation, the constant term is λ1 · λ2 , while the coefficient of the first
order term is equal to −(λ1 + λ2 ). In our case λ1 · λ2 = 4 > 0, therefore the eigenvalues
have the same sign. Since λ1 + λ2 = α, we have that

λ1 < 0 and λ2 < 0 if α < −4

λ1 > 0 and λ2 > 0 if α > 4.

We sum up all the results in the following table:

α ≤ −4 λ1,2 ∈R λ1,2 < 0 O asymptotically stable O stable node

α=0 λ1,2 ∈C Re λ1 = Re λ2 = 0 O stable, non-attractive O centre
0<α<4 λ1,2 ∈C Re λ1 = Re λ2 > 0 O completely unstable O unstable focus
α≥4 λ1,2 ∈R λ1,2 > 0 O completely unstable O unstable node

⃝
c 2018 Politecnico di Torino
Systems of linear ordinary differential equations: solved exercises 25

15c) We have to study the stability of the origin as α varies in R of the system
( ) ( ) ( )( )
ẋ x α −1 x
=A = .
ẏ y 1 1 y

The characteristic equation is

λ2 − (α + 1)λ + (α + 1) = 0.

Also in this case we study the sign of the discriminant:

∆ = (α + 1)2 − 4(α + 1) = (α + 1)(α − 3) < 0 is and only if α ∈ (−1, 3).

• If α ∈ (−1, 3), the characteristic equation has two complex conjugate solutions, with
α+1
Re λ1,2 = >0
2
• If α = −1, λ1 = λ2 = 0. The eigenvalue has algebraic multiplicity 2; it’s easily verified
that its geometric multiplicity is 1.

• If α = 3, λ1 = λ2 = 2 > 0.

• If α < −1, λ1 · λ2 = α + 1 < 0, hence the two eigenvectors have opposite signs.

• If α > 3, λ1 · λ2 = α + 1 > 0 and λ1 + λ2 = α + 1 > 0. Hence λ1 > 0 and λ2 > 0.

We sum up the results in the following table:

α < −1 λ1,2 ∈R λ1 < 0, λ2 > 0 O unstable, non-repulsive O saddle

α = −1 λ1,2 ∈R λ1,2 = 0, geom.mult.1 O completely unstable line of critical points
−1 < α < 3 λ1,2 ∈C Re λ1 = Re λ2 > 0 O completely unstable O unstable focus
α≥3 λ1,2 ∈R λ1,2 > 0 O completely unstable O unstable node

Exercise 16.

16a) We have to study the stability of the origin as α varies in R of the system
( ) ( ) ( )( )
ẋ x 1 α2 x
=A = 1 .
ẏ y α α y

The characteristic equation is

λ2 − (1 + α)λ = λ[λ − (1 + α)] = 0 ⇐⇒ λ1 = 0 ∨ λ2 = 1 + α.

⃝
c 2018 Politecnico di Torino
26 Systems of linear ordinary differential equations: solved exercises

λ2 < 0 if and only if α < −1. When α = −1, both eigenvalues are zero; λ = 0 has geometric
multiplicity 1 since ( )
1 1
A − λI = A = .
1 1

In any case, the system has a line (parallel to the eigenvector associated to the eigenvalue
λ = 0) of critical points. We sum up the results in the following table:

α < −1 λ1 = 0, λ2 < 0 O stable, non-attractive

α = −1 λ1,2 = 0, geom.mult. 1 O unstable, non-repulsive
α > −1, α ̸= 0 λ1 = 0, λ2 > 0 O unstable, non-repulsive

16b) If α ̸= −1, and ui is the eigenvector associated to λi (i = 1, 2), the general solution is such as

c1 u1 + c2 e(1+α)t u2 , c1 , c2 ∈ R;

While, if α = −1 and v is a generalized eigenvector, the general solutions is

c1 u1 + c2 tv, c1 , c2 ∈ R.

Therefore, there exists no α ∈ R such that all the solutions are bounded in R; actually, for
all α ∈ R, the only solutions bounded in R are those corresponding to c2 = 0, that is the
constant solutions.

16c) Using the previous results, we conclude that:

• if α < −1, all the solutions are bounded in [0, +∞);

• if α ≥ −1, the bounded solutions are the constant solutions, that is those corresponding
to c2 = 0.

Exercise 17.

17a) We have to study the stability of the origin as α varies in R of the system
( ) ( ) ( )( )
ẋ x α + 1 −α2 − 1 x
=A = .
ẏ y 1 α y

The characteristic equation is

λ2 − (2α + 1)λ + (2α2 + α + 1) = 0.

⃝
c 2018 Politecnico di Torino
Systems of linear ordinary differential equations: solved exercises 27

Observe that

∆ = (2α + 1)2 − 4(2α2 + α + 1) = −4α2 − 3 < 0, ∀α ∈ R.

2α + 1
Hence the eigenvalues of A are complex conjugate, with Re λ1,2 = . Therefore:
2

α < − 12 Re λ1,2 < 0 O asymptotically stable O stable focus

α = − 12 Re λ1,2 = 0 O asymptotically stable, non-attractive O centre
α > − 12 Re λ1,2 > 0 O completely unstable O unstable focus

17b) Since the eigenvalues are complex conjugate, there exist complex conjugate eigenvectors u±iv.
The general solution is

etRe λ1 [c1 (u cos t − v sin t) + c2 (u sin t + v cos t)] , c 1 , c2 ∈ R.

1
If α = − , the origin is a centre: all the solutions are periodic, therefore bounded in R.
2
1
If α ̸= − , no non-trivial solution is bounded in R.
2
1
17c) The solutions are bounded in [0, +∞) if Re λ1 ≤ 0, that is if α < − .
2

⃝
c 2018 Politecnico di Torino
28 Systems of linear ordinary differential equations: solved exercises

Exercise 18.

18a) We have to study the stability of the origin as α varies in R of the system
( ) ( ) ( )( )
ẋ x α+2 1 x
=A = .
ẏ y −1 α y
The characteristic equation is

λ2 − 2(α + 1)λ + (α2 + 2α + 1) = [λ − (α + 1)]2 = 0.

For all α ∈ R, the matrix A has a unique eigenvalue λ = α + 1, with algebraic multiplicity 2.
Moreover, for all α ∈ R, we have
( )
1 1
A − λI = A − (α + 1)I =
−1 −1
Since A − λI has rank 1, the eigenvalue has geometric multiplicity 1. Moreover, λ = α + 1 < 0
if and only if α < −1.

Therefore, we get:

α < −1 λ1 = λ2 < 0 O asymptotically stable O stable node

α = −1 λ1,2 = 0, molt. geom. 1 O unstable, non-repulsive line of critical points
α > −1 λ1 = λ2 > 0 O completely unstable O unstable node

18b) If we denote by u an eigenvector associated to λ and v a generalized eigenvector, the general

solution is
c1 eλt u + c2 eλt [tu + v], c1 , c2 ∈ R.

Therefore:

• If α < −1, all the solutions are bounded in the future; no non-trivial solution is bounded
in the past.

• If α = −1, the solutions with c2 ̸= 0 aren’t bounded, neither in the future nor in the
past. If c2 = 0, we have ∞1 constant solutions, which are bounded both in the future
and in the past.

• If α > −1, the are no non-trivial solutions bounded in the future. All the solutions are
bounded in the past.

18c) For α = 1, we have the system

( ) ( ) ( )( )
ẋ x 3 1 x
=A = .
ẏ y −1 1 y

⃝
c 2018 Politecnico di Torino
Systems of linear ordinary differential equations: solved exercises 29

) (
a
The eigenvalue is λ = 2. We determine an eigenvector u = associated to the eigenvalue
b
2: ( )( ) ( )
1 1 a 0
(A − λI)u = =
−1 −1 b 0
which corresponds to the equation a + b = 0. Thus we get an eigenvector
( )
1
u= .
−1
( )
a
A vector v = is a generalized eigenvector if and only if it satisfies the linear system
b
( )( ) ( )
1 1 a 1
(A − λI)u = =
−1 −1 b −1
which corresponds to the equation a + b = 1. If we choose b = 0, we get

( )
1
v= .
0
The general solution of the system is
( ) [ ( ) ( )]
1 1 1
c1 e2t 2t
+ c2 e t + c 1 , c2 ∈ R,
−1 −1 0
or, equivalently,
{
x(t) = c1 e2t + c2 e2t (t + 1) = e2t (c1 + c2 + c2 t)
c1 , c2 ∈ R.
y(t) = −c1 e2t − c2 te2t = e2t (−c1 − c2 t),

Exercise 19.

19a) We have to determine the general solution of the system

( ) ( ) ( )( )
ẋ x 1 0 x
=A = .
ẏ y −2 −1 y
The matrix A is triangular, therefore its eigenvalues are its entries on the principal diagonal,
that is λ1 = 1 and λ2 = −1.
( )
a
• λ1 = 1 If we set u = , we obtain the linear system
b
( )( ) ( )
0 0 a 0
= ,
−2 −2 b 0

⃝
c 2018 Politecnico di Torino
30 Systems of linear ordinary differential equations: solved exercises

which corresponds to the equation −2a − 2b = 0. Hence an eigenvector associated to

the eigenvalue λ1 = 1 is ( )
1
u1 = .
−1

• λ2 = −1 We obtain the linear system

( )( ) ( )
2 0 a 0
= ,
−2 0 b 0

which corresponds to the equation 2a = 0. Hence an eigenvector associated to the

eigenvalue λ2 = −1 is ( )
0
u2 = .
1

The general solution of the system is

( ) ( ) ( )
x(t) 1 −t 0
= c1 e t
+ c2 e , c1 , c2 ∈ R,
y(t) −1 1

which can be written as,

{
x(t) = c1 et
c1 , c2 ∈ R.
y(t) = −c1 et + c2 e−t

19b) Since λ1 = 1 > 0 and λ2 = −1 < 0, the zero solution is unstable, non-repulsive; the origin is
a saddle.

19c) If we set c1 = 0 we obtain the solutions bounded in the future (and unbounded in the past);
if we set c2 = 0 we obtain the solutions bounded in the past (and unbounded in the future).
There are no non-trivial solutions bounded in R.

19d) We have to find the real numbers α such that, for all c1 , c2 ∈ R:
 
( ) c1 e(α+1)t ( )
αt x(t)   0
lim e = lim = .
t→+∞ y(t) t→+∞ 0
− c1 e(α+1)t + c2 e(α−1)t

This happens if the wollowing inequalities hold:

{
α−1<0
α + 1 < 0.

Hence, all the solutions satisfy the required condition if and only if α < −1.

⃝
c 2018 Politecnico di Torino