Mathematical Models for Dynamical Systems

N. Boughiout

2025
ii
Contents

1 Introduction 1
1.1 Discrete Dynamical Systems . . . . . . . . . . . . . . . . . . . 2
1.2 continuous dynamical system . . . . . . . . . . . . . . . . . . 3
1.3 Cauchy problem . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Linear Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4.1 vector field . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4.2 diagonalization . . . . . . . . . . . . . . . . . . . . . . 6
1.4.3 exponentials of Matrix . . . . . . . . . . . . . . . . . . 8
1.4.4 The Fundamental Theorem for Linear Systems . . . . . 8
1.4.5 Complex Eigenvalues . . . . . . . . . . . . . . . . . . . 8
1.4.6 Multiple Eigenvalues . . . . . . . . . . . . . . . . . . . 9
1.5 Non-Linear Systems . . . . . . . . . . . . . . . . . . . . . . . . 10
1.6 Stability of Dynamical Systems and Lyapunov’s Theorem . . . 10
1.6.1 Equilibria . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.6.2 Linear stability analysis . . . . . . . . . . . . . . . . . 11
1.6.3 Nonlinear stability analysis . . . . . . . . . . . . . . . . 14
1.7 Exercices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

iii
iv CONTENTS
Chapter 1

Introduction

In mathematics, chemistry, or physics, a dynamical system is a system whose

evolution over time is described by a law. This may refer, for example, to
the progress of a chemical reaction over time, the motion of planets in the
solar system (governed by Newton’s universal law of gravitation), or the evo-
lution of a computer’s memory under the action of a program. Formally, one
distinguishes between discrete-time dynamical systems (such as a computer
program,...) and continuous-time dynamical systems (such as a chemical re-
action or a motion of planets,...).

A dynamical system is said to be deterministic if, for a given initial con-

dition at the present time, there exists exactly one possible future state at
each subsequent instant. This is typically the case for systems modeled by
ordinary differential equations, provided that the conditions for existence and
uniqueness of solutions are satisfied.
Conversely, a system is said to be stochastic if its evolution is governed by
probabilistic laws. Such systems are often described by stochastic differential
equations.

Definition 1. A dynamical system is a pair (X, U ), where X is the state

space, U is a mapping that satisfies :

U : R+ × X → X (1.1)
such that

• U (., x) : R+ → X is continuous.

• U (t, .) : X → X is continuous.

• U (0, x) = x

1
2 CHAPTER 1. INTRODUCTION

• U (t + s, x) = U (t, U (s, x))

Definition 2. A state space is the set of all state variables that are necessary
and sufficient to completely describe the evolution of a dynamical system at
any given time.

1.1 Discrete Dynamical Systems

A discrete dynamical system (sometimes called a time-discrete system) is
generally defined by a recurrence relation :

U n+1 = U n ◦ U (1.2)

for each n and m ∈ N, we have

U n+m = U n ◦ U m (1.3)

Note that U 0 = Id, whene Id is the identity map. If the mapping U is

invertible, we also define U −n = (U −1 )n . In this case, the identity (1.3) is
satisfied for all n, m ∈ Z.

Example 1.
xn+1 = r xn (1 − xn ), n = 0, 1, 2, . . . (1.4)

This is called the logistic map, a famous discrete dynamical system of the
model (population dynamics). Phase space is the interval [0, 1], Parameter:
0 < r ≤ 4, and the behavior depending on the value of r, the system may
converge to a fixed point, oscillate periodically, or exhibit chaotic dynamics.
for example if r = 2.5 and x0 = 0.2 : x1 = 0.4, x2 = 0.6 and x3 = 0.6, so the
sequence quickly stabilizes at a fixed point x = 0.6

Example 2.
xn+1 = a xn + b (1.5)

- Phase space: R.
- This is a simple discrete system that models growth, decay, or convergence
depending on the parameter a.

Both are standard textbook examples of discrete-time dynamical systems.

1.2. CONTINUOUS DYNAMICAL SYSTEM 3

1.2 continuous dynamical system

A continuous-time dynamical system is defined by an ordinary differential
equation (or a system of equations) that describes the evolution of a state
x(t) as a function of time t.
Hence, a continuous dynamical system is defined by
dx
= f (x(t)), x(0) = x0 (1.6)
dt
where x(t) ∈ Rn is the state of the system t, f : Rn → Rn is a vector field
describing the dynamics, and x0 is the initial condition.
The trajectories of the system are continuous functions of time and de-
scribe how the state changes according to the law given by f .
Example 3. A classic example of a continuous dynamical system is the
harmonic oscillator (such as an ideal spring).
(
ẋ(t) = v(t),
k
(1.7)
v̇(t) = − m x(t),

where x(t) denotes the position, v(t) the velocity, m the mass, and k the
spring constant.
Example 4. The two-body problem models the motion of two masses m1 and
m2 interacting only through their mutual gravitational attraction.
From Newton’s law, the relative motion satisfies the second-order equation:
µ
ẍ(t) = − x(t),
∥x(t)∥3
where G is the gravitational constant and µ is the gravitational parameter.
To rewrite the problem as a continuous dynamical system, we introduce the
velocity
v(t) = ẋ(t).
Then the system becomes

ẋ(t) = v(t),
µ
v̇(t) = − x(t).
∥x(t)∥3
Defining the state vector  
x(t)
X(t) = ,
v(t)
4 CHAPTER 1. INTRODUCTION

we can write the system in compact form as

!
v(t)
Ẋ(t) = µ
− x(t) .
∥x(t)∥3

1.3 Cauchy problem

Given an initial condition x0 representing the complete state of the physical
system in its phase space at an initial time t0 , determine the complete state
of the system x(t) in its phase space for any later time t > t0 . The solu-
tion to this fundamental problem lies in the Cauchy–Lipschitz theorem,
which guarantees (under fairly general assumptions) the local existence and
uniqueness of the solution of a differential equation.

Definition 3. Let
f : D ⊂ R × Rn → Rn
be a function defined on a domain D. We say that f is Lipschitz contin-
uous in x on D if there exists a constant L > 0 such that

∥f (t, x1 ) − f (t, x2 )∥ ≤ L∥x1 − x2 ∥, (1.8)

for all (t, x1 ), (t, x2 ) ∈ D with the same first component t.

Theorem 1. Let D ⊆ R × Rn be a closed rectangle with (t0 , x0 ) ∈ int(D)

the interior of D. Let f : D → Rn be a function that is continuous in t
and Lipschitz continuous in x (with Lipschitz constant independent from t).
Then there exists some ϵ > 0 such that the initial value problem
dx
= f (x(t)), x(0) = x0 (1.9)
dt
has a unique solution x(t) in the interval [t0 − ϵ, t0 + ϵ].

Theorem 2 (Existence of solutions without Lipschitz condition). Let (1.9),

where f : D ⊂ R × Rn → Rn is a continuous function on D. Then, for any
point (t0 , x0 ) ∈ D, there exists at least one local solution x(t) defined on some
interval I containing t0 .

Remark 1. This result is known as the Cauchy–Peano theorem. It guar-

antees the existence of solutions under the sole assumption of continuity of f ,
but uniqueness is not guaranteed. Indeed, if f fails to satisfy a Lipschitz
condition in x, multiple solutions may exist.
1.4. LINEAR SYSTEMS 5

Example 5. Consider the differential equation

p
ẋ = |x|, x(0) = 0.
p
The function f (x) = |x| is continuous but not Lipschitz near x = 0. This
equation admits many solutions.

1.4 Linear Systems

Definition 4. A linear dynamical system is a system in which the dy-
namics are governed by linear differential equations (in continuous time) or
linear difference equations (in discrete time).
In continuous time:

Ẋ = AX(t) + B, X(0) = X0 , (1.10)

whene X(t) is the state vctor,A ∈ Rn×n is the evolution matrix and B is a
constant vector in R⋉ .
In discret time :
Xk+1 = AXk + B, X0 given (1.11)
with the same notations.
If B = 0, then the system is said to be homogeneous. The methode of
separation of variables can be used to solve the first-order linear differential
equation in R
ẋ(t) = ax. (1.12)
The general solution is given by

x(t) = x0 eat (1.13)

Example 6. We consider the uncoupled linear system

(
x˙1 = −x1 ,
(1.14)
x˙2 = 2x2 ,

The general solution is given by

(
x1 = c1 e−t ,
(1.15)
x2 = c2 e2t ,

or equivalently by  −t 
e 0
X(t) = · C, (1.16)
0 e2t
6 CHAPTER 1. INTRODUCTION
 
c
where C = 1 . The dynamical system for this example is given by
c2

e−t 0
 
Φ(t, C) = · C,
0 e2t

1.4.1 vector field

Geometrically, the dynmical system describes the motion of the points in
phase space along the solution curves defined by the system of differential
equations.
The function
f (X) = AX, (1.17)
on the right-hand side of (1.11) defines a mapping f : Rn → Rn (linear in
this case). this mapping defines a vector field on Rn . At each point in phase
space, the vector field “tells” the system where to go next. If we draw arrows
f (x) at every point x we get a phase portrait: a geometric representation of
trajectories.

Example 7.  
0 −1
ẋ(t) = Ax(t), A= ,
1 0
Here the vector field is:  
−x2
f (x) = .
x1
which generates circular trajectories (rotation).

1.4.2 diagonalization
The algebraic technique of diagonalizing a square matrix A can be used to
reduce the linear system
Ẋ = AX (1.18)
The following theorem from linear algebra then allows us to solve the above
linear system.

Theorem 3. If the eigenvalus λ1 , λ2 , ..., λn of an n × n matrix A are real

and distinct, then any set of corresponding eigenvectors v1 , v2 , ..., vn forms a
basis for Rn , the matrix P = [v1 v2 ...vn ] is invertible and

P −1 AP = diag[λ1 , ..., λn ]
1.4. LINEAR SYSTEMS 7

Corollary 1. Under the hypothesis of the above theorem, the solution of the
linear system (1.11) is given by
X(t) = P E(t)P −1 X(0),
where E(t) is is the diagonal matix
E(t) = diag[eλ1 t , ..., eλn t ]
.
Example 8. Consider the linear system
(
ẋ1 = −x1 − 3x2
ẋ2 = 2x2
which can be written in the form (1.11) with the matrix
 
−1 −3
A=
0 2
the eigenvalues of A are λ1 = −1 and λ2 = 2, a pair of corresponding
eigenvectors is given by
   
1 −1
v1 = , v2 =
0 1
The matrix P is its inverse are thne given by
   
1 −1 −1 1 1
P = , and P =
0 1 0 1
We verify that  
−1 −1 0
P AP = .
0 2
Thene under the coordinate transformation Y = P −1 X, we obtain the un-
coupled linear system
ẏ1 = −y1 ẏ2 = 2y2
which has the general solution y1 (t) = c1 e−t, y2 (t) = c2 e2t . And according
to the above corollary, the general solution to the original system is given by
 −t 
e 0
X(t) = P P −1 C
0 e2t
where C = X(0), or equivalenty by
x1 (t) = c1 e−t + c2 (e−t − e2t )x2 (t) = c2 e2t
.
8 CHAPTER 1. INTRODUCTION

1.4.3 exponentials of Matrix

Definition 5. Let A be an n × n matrix thene for t ∈ R,
inf ty
X Ak tk
eAt = .
k=0
k!

For an n × n matrix A, eAt is an n × n matrix which can be computed in

terms of the eigenvalues and eigenvectors of A.
Corollary 2. • If P −1 AP = diag[λj ], then eAt = P diag[eλj t]P −1
• eS+T = eS eT
   
a −b a cos b − sin b
• If A = A
then e = e
b a sin b cos b

1.4.4 The Fundamental Theorem for Linear Systems

p 16
Lemma 1. Let A be a square matrix, then
d At
e = AeAt
dt
Proof. We have
Theorem 4. (The Fundamental Theorem for Linear Systems)
Let A be an n × n matrix. Thene for a given x0 ∈ Rn , the initial value
problem
ẋ = Ax
(1.19)
x(0) = x0
has a unique solution given by

x(t) = eAt x0 .

1.4.5 Complex Eigenvalues

Theorem 5. If the 2n×2n real matrix A has 2n distinct complex eigenvalues
¯ j = aj − ibj and corresponding eigenvectors wj =
λj = aj + ibj and lambda n
uj + ivj and w̄j = uj − ivj , j = 1, ..., n, then u1 , v1 , ..., un , vn is a basis
for R2 n, the matrix
 
P = v1 u1 v2 u2 . . . vn un
1.4. LINEAR SYSTEMS 9

is invertible and  
−1 aj b j
P AP = diag
−bj aj
a real 2n × 2n matrix with 2 × 2 blocks along the diagonal.
Corollary 3. Under the hypotheses of the above theorem, the solution of the
initial value problem (1.18) is given by
 
aj t cos bj t − sin bj t
x(t) = P diage P −1 x0 (1.20)
sin bj t cosbj t

Example 9. Solve the initial value problems :

 
1 −1 0 0  
1 1 −3 0 0
0 0 (2)A =  0 3 −2
(1)A = 0 0 −3 2
0 1 1
0 0 1 1

1.4.6 Multiple Eigenvalues

Definition 6. Let λ be a eigenvalue of the n × n matrix A of multiplicity
m ≤ n. then for k = 1, ..., m, any nonzero solution v of

(A − λI)k v = 0 (1.21)

is called a generalized eigenvector of A.

Definition 7. An n×n matrix N is said to be nilpotent of order k if N k−1 ̸= 0
and N k = 0.
Corollary 4. Under the hypotheses of the above theorem, the linear system
(1.18) has the solution

N k−1 tk−1
x(t) = P diag[eλj t ]P −1 [I + N t + ... − ]x0 (1.22)
(k − 1)!
Example 10. Solve the initial value problem (1.18) with
 
3 1
(1)A =
−1 1
 
0 −2 −1 −1
1 2 1 1
(2)A =  0 1

1 0
0 0 0 1
10 CHAPTER 1. INTRODUCTION

1.5 Non-Linear Systems

A nonlinear dynamical system is a system whose evolution over time is gov-
erned by nonlinear equations. Mathematically, it can be written as

ẋ(t) = f (x(t), t), x(0) = x0 , (1.23)

where x(t) ∈ Rn is the state vector and f : Rn × R → Rn is a nonlinear

function in x.

Definition 8. A dynamical system is said to be autonomous if its evolution

depends only on the current state and not explicitly on time. Formally, it
can be written as:
ẋ(t) = f (x(t)) (1.24)

Example 11. A well-known example of a first-order nonlinear dynamical

system is the simple pendulum. Let θ(t) denote the angular displacement and
ω(t) = θ̇(t) the angular velocity. The equations of motion are given by
(
θ̇(t) = ω(t),
g
ω̇(t) = − sin(θ(t)),
L
where g is the gravitational acceleration and L is the length of the pendulum.

The resolution of nonlinear dynamical systems is a central topic in ap-

plied mathematics, physics, and engineering. Unlike linear systems, for which
general analytical methods are available, nonlinear systems are often signif-
icantly more complex. Exact analytical solutions are very rare in practice
and can only be obtained for relatively simple cases (e.g., separable equa-
tions, substitutions, first integrals, symmetries, or invariants). For this rea-
son, qualitative analysis and numerical methods are generally preferred. Lin-
earization provides a useful first approximation of the system’s behavior in
the neighborhood of an equilibrium point.

1.6 Stability of Dynamical Systems and Lya-

punov’s Theorem
In the study of dynamical systems, stability refers to the behavior of solutions
under small perturbations of the initial conditions.
1.6. STABILITY OF DYNAMICAL SYSTEMS AND LYAPUNOV’S THEOREM11

1.6.1 Equilibria
Definition 9. In a dynamical system defined by the equation (1.23), an
equilibrium point x∗ is a point where f (x∗ ) = 0.

Remark 2. An equilibrium point is often referred to as a singular point,

fixed point, or critical point.

Remark 3. If the point x∗ is an equilibrium point for the system (1.23) then,
with the change of variables y = x − x∗ , (1.9) can be rewritten as

ẏ = Ay (1.25)

Notice that y = 0 is an equilibrium point for the equation (1.24).

1.6.2 Linear stability analysis

Definition 10. We define the stability of the equilibrium point x∗ as follows

• The equilibrium point x∗ is stable if,

∀ϵ > 0, ∃δ(ϵ) > 0 : ∥x(0) − x∗ ∥ < δ ⇒ ∥x(t) − x∗ ∥ < ϵ, ∀t ≥ 0.

(1.26)

• The equilibrium point x∗ is asymptotically stable if it is stable and,

∥x(t) − x∗ ∥ −→ 0 as t → ∞.

• The equilibrium point x∗ is unstable if it is not stable.

Remark 4. Asymptotic stability indicates that the trajectories of the system

converge to the equilibrium point as t → ∞ . In contrast, neutral stability
(stable but not asymptotically stable) means that while the trajectories remain
near the equilibrium point, they do not necessarily converge to it over time.

Consider the linear system (1.24), where A is a constant n×n matrix, and
let y = 0 be an equilibrium point. The behavior near an equilibrium point
(at the origin in this case) is determined by the eigenvalues of A. Specifically,
the characteristic equation det(A − λI) = 0 gives the eigenvalues λi of A ,for
i = 1, , n.

Theorem 6. • If all eigenvalues of A have negative real parts, then the

equilibrium y = 0 is asymptotically stable.

• at least one eigenvalue of A has a positive real part, then the equilibrium
y = 0 is unstable.
12 CHAPTER 1. INTRODUCTION

• If all eigenvalues of A have non-positive real parts and at least one

eigenvalue has a real part of zero, the stability of y = 0 cannot be de-
termined solely from the eigenvalues of A and requires further analysis.
Proof. The general solution to the system ẏ = Ay is given by y(t) = eAt y(0).
To analyze the stability of y = 0, we examine the behavior of ∥y(t)∥ as
t → ∞. The properties of eAt depend on the eigenvalues of A.
• Case 1: All eigenvalues of A have negative real parts.
If all eigenvalues of A have negative real parts, then eAt decays to zero
as t → ∞. Specifically, since the real parts of the eigenvalues determine
the rate of decay, we have:

∥eAt ≤ Ce−αt |

for some constants C > 0 and α > 0 (where α is related to the smallest
magnitude of the negative real parts of the eigenvalues). It follows that

∥y(t)∥ = ∥eAt y(0)∥ ≤ Ce−αt ∥y(0)∥ −→ 0, as t → ∞.

Thus, y = 0 is asymptotically stable.

• Case 2: At least one eigenvalue of A has a positive real part.

If A has at least one eigenvalue with a positive real part, there exists a
component of the solution that grows exponentially as t → ∞.
Let λ be an eigenvalue of A with Re(λ) > 0, and let v be the cor-
responding eigenvector. Then the term eλt v appears as part of the
solution y(t), resulting in

∥y(t)∥ → ∞ as t → ∞.

Hence, y = 0 is unstable.

• Case 3: All eigenvalues of A have non-positive real parts, with at least

one zero real part.
If all eigenvalues have non-positive real parts and at least one eigenvalue
has zero real part, the stability analysis is more subtle. The matrix A
may have both decaying and non-decaying modes. In this case:

– If A is diagonalizable, the solution can be expressed as a combina-

tion of terms that either decay (from eigenvalues with negative real
parts) or remain constant (from eigenvalues with zero real parts).
Thus, y = 0 is stable but not asymptotically stable, as certain
components of y(t) do not decay to zero but remain bounded.
1.6. STABILITY OF DYNAMICAL SYSTEMS AND LYAPUNOV’S THEOREM13

– If A is not diagonalizable, the matrix eAt may exhibit terms that

grow linearly in t, due to terms involving teλt when λ = 0. In
this situation, the solution y(t) can grow unboundedly as t → ∞,
rendering y = 0 unstable. This completes the proof.

Remark 5. To analyze the stability of the equilibrium point x∗ in the original

system ẋ = Ax + B, we equivalently study the stability of y = 0 in the shifted
system ẏ = Ay, where y = x − x∗ . In this approach, by shifting the system
to center around the equilibrium x∗ , we reduce the analysis to examining the
stability of the zero equilibrium in the transformed system, simplifying the
calculations.

Classification of fixed points in 2D

Consider the linear differential system (1.23), where A is a 2 × 2 matrix. The
nature of fixed points (equilibria) can be classified based on the eigenvalues
of the matrix A. The eigenvalues λ1 and λ2 of A can lead to different types
of fixed points:

• Nodes: Both eigenvalues are real and have the same sign i.e. λ1 λ2 > 0
and Im(λ1,2 ). If negative, the node is a stable (or attracting) node; if
positive, it is an unstable (or repelling) node.

• Saddles: Eigenvalues are real with opposite signs i.e. λ1 λ2 < 0 and
Im(λ1,2 ) = 0. Saddles are always unstable.

• Spirals (Foci): Eigenvalues are complex conjugates with non-zero real

parts i.e. Re(λ1,2 ) ̸= 0. If the real parts are negative, it is a stable
spiral (focus); if positive, it is an unstable spiral.

• Centers: Eigenvalues are purely imaginary i.e. Re(λ1,2 ) = 0, leading

to closed orbits around the equilibrium. Centers are typically stable
but not asymptotically stable.

Example 12. Consider the dynamical system given by:

 
−2 1
ẋ = x
1 −3

To analyze the stability of the equilibrium point at the origin, we first de-
termine the characteristic equation. This is achieved by computing the de-
terminant of the matrix obtained by subtracting λ (the eigenvalue) from the
14 CHAPTER 1. INTRODUCTION

diagonal entries:  
−2 − λ 1
det =0
1 −3 − λ
This leads to the eigenvalues:
λ1 = −1, and λ2 = −4
Since both eigenvalues are negative λ1 < 0 and λ2 < 0, we conclude that the
equilibrium point at the origin is a stable node.

1.6.3 Nonlinear stability analysis

Consider the nonlinear differential system
ẋ = f (x) (1.27)
where x ∈ Rn and f : Rn → Rn is a sufficiently smooth function.
Definition 11 (Flow of a nonlinear differential system). Consider the non-
linear differential system (1.27)
The flow associated with system is a function
φ : D ⊂ R × Rn → Rn ,
that assigns to each initial condition x0 and time t the unique solution φ(t, x0 )
of(1.27) satisfying the initial condition
φ(0, x0 ) = x0 .
If the solution exists for all t in some interval containing zero, the function
φ(t, x0 ) describes the trajectory of the system starting from x0 . The flow
satisfies the semigroup property:
φ(t + s, x0 ) = φ(t, φ(s, x0 )),
for all t, s where the flow is defined.
Definition 12 (Hyperbolic equilibrium point). An equilibrium point x0 is
called a hyperbolic equilibrium point of system (1.27) if none of the eigen-
values of the matrix Df (x0 ) have zero real part, where Df (x0 ) is the n × n
Jacobian matrix of (1.27) at x0 .
Definition 13 (Linearized system). The linear system
ẋ = Ax, (1.28)
with the matrix A = Df (x0 ) is called the linearized system of (1.27) at x0 .
1.6. STABILITY OF DYNAMICAL SYSTEMS AND LYAPUNOV’S THEOREM15

Definition 14. Two autonomous systems of differential equations, such as

(1.27) and (1.28), are said to be topologically equivalent in a neighborhood
of the origin, or to have the same qualitative structure near the origin, if there
exists a homeomorphism H mapping an open set U containing the origin onto
an open set V containing the origin, such that H maps trajectories of (1.27)
in U onto trajectories of (1.28) in V while preserving their orientation in
time. That is, if a trajectory is directed from x1 to x2 in U , then its image
is directed from H(x1 ) to H(x2 ) in V .
If, in addition, the homeomorphism H preserves the parameterization
by time, then the systems (1.27) and (1.28) are said to be topologically
conjugate in a neighborhood of the origin.
In the following, we present a simple example of two topologically conju-
gate linear systems.
Example 13. Consider the linear systems

ẋ = Ax, ẏ = By, (1.29)

where the matrices A and B are given by

! !
−1 −3 2 0
A= , B= .
−3 −1 0 −4

Let H(x) = Rx, where the matrix R is

! !
1 1 1 1 1 −1
R= √ , R−1 = √ .
2 −1 1 2 1 1

It follows that B = RAR−1 , and setting y = H(x) = Rx, equivalently

x = R−1 y, we obtain
ẏ = RAR−1 y = By.
Thus, if x(t) = eAt x0 is the solution of the first system through x0 , then

y(t) = H(x(t)) = Rx(t) = ReAt x0 = eBt Rx0 ,

is the solution of the second system through Rx0 . That is, H maps trajectories
of the first system onto trajectories of the second system while preserving their
parameterization, since
HeAt = eBt H.
The mapping H(x) = Rx corresponds to a rotation by 45◦ and is clearly
a homeomorphism.
16 CHAPTER 1. INTRODUCTION

The following theorem is a fundamental result in the local qualitative

theory of ordinary differential equations. It establishes that near a hyper-
bolic equilibrium point x0 , the nonlinear system (1.27) exhibits the same
qualitative behavior as the corresponding linear system

ẋ = Ax, (3.2.7)

where A = Df (x0 ). In what follows, we assume that the equilibrium point

x0 has been translated to the origin.

Theorem 7. Let E be an open subset of Rn containing the origin, let f ∈

C 1 (E), and let φt be the flow of the nonlinear system (1.27). Suppose that
f (0) = 0 and that the matrix A = Df (0) has no eigenvalue with zero real
part. Then there exists a homeomorphism H of an open set U containing the
origin onto an open set V containing the origin such that for each x0 ∈ U ,
there exists an open interval I0 ⊂ R containing zero, such that for all x0 ∈ U
and t ∈ I0 ,
H ◦ φt (x0 ) = eAt H(x0 ).

That is, H maps trajectories of the nonlinear system (3.2.4) near the origin
onto trajectories of the linear system (3.2.7) near the origin and preserves
the parameterization by time.

1.7 Exercices
Exercise 1. Prove that the following differential system defines a dynamical
system
ẋ = Ax x0 = x(0)

Where A is a constant matrix, x ∈ Rn and the solution has the form x(t) =
eAtx0 .

Exercise 2. Consider the Cauchy problem

ẋ = f (x) = x cos(x), x(0) = x0 .

Show the existence and the unicity of solution.

Exercise 3. Find the general solution and draw the phase portrait for the
following linear systems :
1.7. EXERCICES 17

ẋ1 = x1
( ( ( ( 
ẋ1 = x1 ẋ1 = x1 ẋ1 = −x2 ẋ1 = −x1 + x2
(1) (2) (3) (4) (5) ẋ2 = x2
ẋ2 = x2 ẋ2 = 2x2 ẋ2 = x1 ẋ2 = −x2 
ẋ3 = x3

 
ẋ1 = −x1
 ẋ1 = −x2

(6) ẋ2 = −x2 (7) ẋ2 = x1
 
ẋ3 = x3 ẋ3 = −x3
 

Exercise 4. Find the eigenvalues and eigenvectors for the matrix A and
show that B = P −1 AP is a diagonal matrix. solve the linear
 system ẏ= By
3 1 1 3
and solve ẋ = Ax using the above corollary. (1) A = (2) A =
  1 3 3 1
  1 0 0
−1 1
(3) A = (4) A = 1 2 0 

1 −1
1 0 −1

Exercise 5. Write the following linear differential equations with constant

coefficients in the form of the linear system and solve : (1) ẍ + ẋ − 2x = 0
(2) ẍ + x = 0
...
(3) x − 2ẍ − ẋ + 2x = 0
Hint: Let x1 = x, x2 = ẋ, etc

Exercise 6. Prove the corollary (2) .

Exercise
  7. Compute the exponentials
 of the folllowing matrices
  : (1) A =
2 1 1 2 1 0 5 −6
(2) A = (3) A = (4) A = (5) A =
0 −3 0 −1  5 1  3 −4 
    1 0 0 1 0 0
2 −1 0 1
(6) A = (7) A = 1 2
 0 (8) A = 0 2 1 (9) A =
1 2 1 0
  0 0 3 0 0 2
2 0 0
1 2 0
0 1 2

Exercise 8. Suppose that the square matrix A has a negative eigenvalue.

Show that the linear system ẋ = Ax has at least one nontrivial solution x(t)
that satisfies
lim x(t) = 0.
t→∞

Exercise 9. Solve the initial value problem

ẋ = Ax
18 CHAPTER 1. INTRODUCTION
 
1
x(0) =
0
for  
−2 −1
A=
1 −2
and sketch the solution curve in phase plane R2 .