1 Introduction

1.1 Differential equations

Let us consider a real function of one real variable y = y (x ). The following equations fall
under the category of ODEs:

y′ = y, y ′′ − y = 0. y ′′ − y 2 = 0.

The solution to the first equation is given by y = ce x , where c represents an arbitrary

constant. To pick out a specific solution, we would have to introduce an initial condition into
the ODE. For instance, if we stipulate that y (0) = 1, then c becomes equal to 1 and the
solution would be y (x ) = e x .
The second equation is a second-order linear ODE. The standard form of such equations
is expressed with three known functions, p, q and r , such that:

y ′′ + p (x )y ′ + q (x )y = r (x ).

Then the ODE is said to be homogeneous if r = 0 and inhomogeneous otherwise. The total
solution y of the ODE 1.1 is the sum of a homogeneous solution yh (solution of the ODE
considering r = 0) and a particular solution yp (y = yh + yp ). The particular solution yp can
be obtained, for instance, using the variation of the constant.
A linear homogeneous second-order ODE with constant coefficients, a,b,c ∈ R, can be
written as
ay ′′ + by ′ + cy = 0.
The solution of such equation depends on those of the characteristic equation

aλ2 + bλ + c = 0.

We distinguish the following cases.

1. Two distinct real roots λ1 and λ2 . The solution is yh = C1 e λ1 x + C2 e λ2 x , C1 , C2 ∈ R.

b
2. Double real root λ = − . The solution is yh = [C1 + C2 x ]e λx .
2a
3. Complex conjugate roots λ1 = α + iβ, λ2 = α − iβ.

yh = e αx [C1 cos(βx ) + C2 sin(βx )] .

3
1.2. BASIC CONCEPTS CHAPTER 1. INTRODUCTION

In contrast, Partial Differential Equations (PDEs) encompass partial derivatives due to

their involvement with more than one independent variable, such as x, y , and others.
Notations
We commonly represent the partial derivatives of a function u with respect to independent
variables x, y , using various notations, including:

∂u ∂ 2u
, , ux , uxy , . . .
∂x ∂y ∂x

For instance, consider a function u = u (x, y ) and the following PDEs:

∂u ∂u ∂u ∂u
= 0, = 0, + = 0.
∂x ∂y ∂x ∂y
Solving the first equation is relatively straightforward, and it can be verified that the solutions
include:
u = c, u = y , and u = y 2 + ln |y |.
In fact, the most general solution takes the form u = f (y ), where f is an arbitrary function.
Similarly, the general solution for the second equation has the form u (x ) = g (x ). The general
solution of the third equation is given by u = f (x − y ), where f is an arbitrary function of
its argument. A nuanced distinction between solutions of ODEs and solutions of PDEs is
that ODE solutions incorporate constants of integration, while PDE solutions encompass
functions of integration.

1.2 Basic concepts

Definition 1.1 (PDE). A partial differential equation (PDE) is an equation involving
one or more partial derivatives of an (unknown) function, that depends on two or more
independent variables. The order of the differential equation is the highest partial derivative
that appears in the equation. More formally, a PDE can be written in the form

F (x1 , x2 , . . . , xn , u, ux1 , . . . , uxn , ux1 x1 , ux1 x2 , . . . ) = 0, (1.1)

where F is a given function of the independent variables x = (x1 , . . . , xn ); of the unknown

function u = u (x) and of a finite number of its partial derivatives. The PDE (1.1) can also
be written in the form
Lu = f , (1.2)
where where f = f (x) and L is an operator that contains all the operations (differentiation,
multiplication, composition, etc.) that act on u. Without loss of generality, we consider
n = 2 in th remaining.

4
CHAPTER 1. INTRODUCTION 1.2. BASIC CONCEPTS

Definition 1.2 (Classification).

• We say that a PDE is linear if L is linear and nonlinear otherwise.

• A PDE is called semilinear when it is linear in the leading (highest-order) terms.

• A PDE of order m is said to be quasi-linear if it is linear in the derivatives of order m

with coefficients that depend on x, u and the derivatives of order < m.

• Linear equations can further be classified as homogeneous when f = 0 and non-

homogeneous otherwise.

Proposition 1.2.1.

• The superposition principle: any linear combination of two solutions of a linear homo-
geneous PDE is also a solution.

• For a linear PDE, the sum of a homogeneous solution and an inhomogeneous solution
is also an inhomogeneous solution.

Example 1.1 (General forms of first order PDEs).

• Linear PDE: a (x, y )ux + b (x, y )uy + c (x, y )u = d (x, y )

• Semi-linear PDE: a (x, y )ux + b (x, y )uy = d (x, y , u )

• Quasi-linear PDE: a (x, y , u )ux + b (x, y , u )uy = d (x, y , u )

Example 1.2.

• The heat equation: ut − uxx = x (linear non-homogeneous second-order, L = ∂t − ∂xx

2 ).

• The heat equation: ∂u

∂t − c 2 ∆u = 0 (second-order, linear, homogeneous, L = ∂t − c 2 ∆).

• The wave equation: utt − uxx − x 3 = 0 (non-homogeneous linear second order, L =

2 − ∂2 .
∂tt xx

∂u 2
• The wave equation: ∂t 2
− c 2 ∆u = 0 (second-order, linear, homogeneous, L = ∂tt
2 −

c 2 ∆).

• The Laplace equation: ∆u = 0 (second-order, linear, homogeneous, L = ∆).

• The Poisson equation: ∆u = f (second-order, linear, non-homogeneous if f ̸= 0).

• The transport (convection) equation: ∂u

∂t ∂x = 0 (first-order, linear, homogeneous)
+ ∂u

• The Burger’s equation: ∂u

∂t ∂x = 0 (first-order, quasilinear, homogeneous, L =
+ u ∂u
∂t + u∂u ).

• The Korteweg de Vries equation (KdV): ut + 6uux + uxxx = 0 (third-order, semi-linear,

homogeneous, L = ∂t + u∂u + ∂xxx3 ).

5
1.3. MODELING WITH PDES CHAPTER 1. INTRODUCTION

Definition 1.3. A solution of a PDE for a function u = u (x, y ) can be visualized as a

surface in the three-dimensional space (x, y , u), called integral surface.

Definition 1.4 (Classification of PDEs problems.). Consider a given PDE on a region Q.

1. Cauchy or initial-value problem: some additional initial-values for the unknown

function and its derivatives are given on some subsets of Q.

2. Boundary value problem: some additional boundary conditions for the unknown
function and its derivatives are given on ∂Q.

3. Mixed type problem: some additional initial-values and boundary conditions for the
unknown function and its derivatives are given.

Definition 1.5. A problem with a PDE is well-posed in a class of functions C, if the

following three conditions are satisfied.

1. there exists a solution in C;

2. the solution is unique;

3. the solution is continuously dependent on the given conditions, e.g., initial-values,

boundary conditions, coefficients, etc.

Example 1.3. Show that u (x, y ) = y 2 + x is the solution of the Cauchy problem on R 2 :
∂u
= 1 and u (0, y ) = y 2 (y ∈ R ).
∂x

1.3 Modeling with PDEs

PDEs are widely recognized as "Mathematical Physics Equations", with applications spanning
diverse fields such as acoustics, optics, elasticity, hydrodynamics, aerodynamics, electromag-
netism, quantum mechanics and chemistry, seismology, chemical kinetics. Furthermore, PDEs
arise in unexpected arenas like economics, financial mathematics, social sciences, and image
processing. In the following, we delve into the origins of specific PDEs.

1.3.1 Conservation Laws

Studying natural processes, we observe two main tendencies: the tendency to achieve a
certain balance between causes and consequences, or the tendency to break this balance.
We usually use some law or principle that expresses such a balance between the so-called
state quantities u and flow quantities q, and their spatial and time changes. Here are some
examples.

• State quantities: density, pressure, temperature, entropy, ...

• Flow quantities: velocity, momentum, tension, heat flux, ...

6
CHAPTER 1. INTRODUCTION 1.3. MODELING WITH PDES

Consider a function u = u (x, t ) representing the density or the concentration of a physical

quantity m (e.g. a mass in a region, heat in a metal bar, traffic on a highway) and q (u ) is its
flux function (the rate of change of m). If we consider a control interval (x1 , x2 ), the integral

Zx2
u (x, t ) dx
x1

gives the amount of m contained inside (x1 , x2 ) at time t. Without sources or sinks, the rate
of change of m inside (x1 , x2 ) is determined by the net flux through the end points:
x2
∂ Z
u (x, t )dx = −q (u (x2 , t )) + q (u (x1 , t )),
∂x x
1

and thus, for smooth functions u and q, we get

Zx2
[ut (x, t )dx + q (u (x, t ))x ]dx = 0,
x1

which, due to the arbitrariness of the interval (x1 , x2 ), implies

ut + q (u )x = 0, x ∈ R, t > 0. (1.3)

Equation 1.3 is a scalar conservation law that gives a link between density and flux. At
this point we have to establish a constitutive relation for q that describes the flux function.
For a higher dimension, q (u ) becomes a vector-valued function. With a scalar source
function f = f (x, t ), the conservation law writes:

ut (x, t ) + ∇· q (u )(x, t ) = f (x, t ), x ∈ RN , t > 0. (1.4)

When reaching a steady or stationary state, all quantities become time-independent and
thus, simplified versions of the conservation laws are obtained.

1.3.2 Constitutive Laws

The conservation law represents one equation for two unknown functions, the state and the
flow quantities. To create a solvable mathematical model, we need another relation between
these functions. Such relations are usually based on the generalization of experimental
observations and depend on the properties of the particular medium or material. A typical
example of the constitutive law from the elasticity theory is Hook’s law, which states that,
for relatively small deformations of an object, the displacement or size of the deformation
(state quantity) is directly proportional to the deforming force or tension (flow quantity).
Some constitutive laws will be given in the next section.

7
1.3. MODELING WITH PDES CHAPTER 1. INTRODUCTION

1.3.3 Basic Models

Advection (Transport) equation

Advection is the transport of a substance such as mass, energy, or concentration, or a conserved

scalar, by a known velocity field (of a fluid). The advection equation is a fundamental equation
used in various fields, including fluid dynamics, meteorology, and transport phenomena. It is
a first-order PDE and can be written in one dimension as:
∂u ∂u
+c = 0,
∂t ∂x

where u (x, t ) is the quantity being transported (e.g., concentration of a substance or a

temperature), t is time, x is the spatial coordinate, and c (x, t ) is the velocity.

Example 1.4. Suppose a certain amount of a chemical spills into a river and flows downstream.
Let us suppose that the concentration is given by u = u (x, t ) and that the river flows with
constant speed c. If we assume that the chemical does not diffuse, then the flux is proportional
to the quantity u: q (x, t ) = cu (x, t ) (which is the constitutive law). Consequently, using
the scalar conservation law, we obtain

ut + cux = 0.

As we will show later, the solution of (1.11) is a function

u (x, t ) = F (x − ct ),

where F is an arbitrary differentiable function. Such a solution is called the right traveling
wave, since its graph at a given time t is the graph of the function F (x ) shifted to the right
by the value ct. Thus, with growing time, the profile F (x ) is moving without changes to the
right at the speed c (see Figure 1.1).

Figure 1.1: Traveling wave

8
CHAPTER 1. INTRODUCTION 1.3. MODELING WITH PDES

Wave equation (vibrating strings). The wave equation is a fundamental PDE that
describes how waves propagate through a medium. It is used to model a wide range of wave
phenomena, including acoustic waves, electromagnetic waves, and mechanical waves. The
one-dimensional wave equation, which represents wave propagation along a single spatial
dimension x is given by:
∂u 2 2 ∂u
2
= c . (1.5)
∂t 2 ∂x 2
Example 1.5. To derive the wave equation in one spacial dimension, we consider an elastic
string with a constant mass density ρ that undergoes small amplitude vertical vibrations. An
infinitesimal element located between x1 and x2 , with ∆x = x2 − x1 is shown in Fig. 1.2.

Figure 1.2: Derivation of the wave equation.

We define u (x, t ) to be the vertical displacement of the string from the x-axis at position
x and time t, T the tension of the string assumed to be constant, and θ the angle between
the string and the horizontal line. We assume that the string element only accelerates
vertically (horizontal forces balance) and that the gravity is negligible compared to T .
We apply the Newton’s law of motion (m x ′′ (t ) = F⃗ ) for the vertical acceleration of our
infinitesimal string element (utt ). The "infinitesimal" mass can be obtained as
q q
m = ρ ∆x 2 + ∆u 2 ≈ ρ∆x 1 + ux2
because ux ≈ ∆u/∆x. Consequently,
q
ρ∆x 1 + ux2 utt = T sin θ2 − T sin θ1 .
With the assumption of small vibrations (∆u << ∆x or ux << 1), we have
q
sin θi ≈ ux (xi , t ), 1 + ux2 ≈ 1.
Therefore, we get
ρ ∆x utt = T (ux (x2 , t ) − ux (x1 , t )).
Dividing by ∆x and taking the limit ∆x → 0 results in the wave equation
utt = c 2 uxx ,
where c 2 = T /ρ.

9
1.3. MODELING WITH PDES CHAPTER 1. INTRODUCTION

Diffusion equation. Diffusion is a fundamental physical process that describes the move-
ment of particles or molecules in a fluid or a solid, from regions of higher concentration to
regions of lower concentration, to spread out and mix with neighboring particles, resulting
in the equalization of the concentration distribution over time (at equilibrium). Diffusion is
governed by the so-called heat equation:
∂u
= ∆u. (1.6)
∂t
Example 1.6. Consider a rod of length L, cross section area A and density ρ(x ) (Fig. 1.3).

Figure 1.3: The typical cross-section

Given the specific heat c (x ) and the temperature u (x, t ), we define the total heat H as
ZL
H (x, t ) = c (x )ρ(x )u (x, t )Adx. (1.7)
0

The amount of thermal energy per unit time flowing to the right per unit surface area is
called “heat flux” defined by Φ(x, t ). If heat is generated within the rod and its heat density
Q(x, t ) is given, then the total heat generated is
ZL
Q(x, t )Adt. (1.8)
0

The conservation of heat energy states that: the rate of change of total heat = heat
flowing across boundaries + heat energy generated inside. Mathematically, we can write
Z L
∂H
= (Φ(0, t ) − Φ(L, t )) A + Q(x, t )Adt. (1.9)
∂t
0

Consequently,
ZL Z Z L L
∂u ∂Φ
c (x )ρ(x ) (x, t )Adx = − Adx + Q(x, t )Adt, (1.10)
∂t ∂x
0 0 0

which gives
ZL !
∂u ∂Φ
c (x )ρ(x ) (x, t ) + − Q(x, t ) Adx = 0, (1.11)
∂t ∂x
0
Since this applies to any length L, then
∂u ∂Φ
c (x )ρ(x ) (x, t ) = − + Q(x, t ) (1.12)
∂t ∂x
10
CHAPTER 1. INTRODUCTION 1.3. MODELING WITH PDES

Fourier suggested the following form for heat flux:

∂u
Φ = −k (1.13)
∂x
where k is the thermal conductivity. This then gives

∂u ∂u 2
= D 2 + f (x, t ), (1.14)
∂t ∂x
where D = k /ρc, is the coefficient of diffusion, and f = Q/ρc, is the source term.

Laplace equation The solutions of Laplace’s equation are used in many important fields of
science, notably the fields of electromagnetism, astronomy, and fluid dynamics, because they
can be used to accurately describe the behavior of electric, gravitational, and fluid potentials.
A function satisfying Laplace’s equation in a region of the plane (or 3-space) is said to be
harmonic.

Example 1.7. Consider the example 1.6 of heat diffusion in 2D case. Suppose that f (x, t )
vanishes after some time. When heat diffuses and reaches an equilibrium (no variations over
time), the heat diffusion equation becomes stationary equations, the simplest of which is
Laplace equation
∆u (x, t ) = 0.

11