National Polytechnic School

Preparatory cycle-ENP Math Analysis2 F.FELLAG

Functions Of Two Variables

Historical Summary of Two-Variable Functions

The study of functions of multiple variables developed gradually:

 Antiquity and Middle Ages: The Greeks (Euclid, Archimedes) studied surfaces and
volumes. Arab mathematicians developed algebra.
 17th century: Descartes introduced analytic geometry, while Newton and Leibniz
developed differential calculus.
 18th and 19th centuries: Euler formalized partial derivatives, Lagrange developed
optimization, and Gauss applied these concepts to physics and geometry.
 20th and 21st centuries: Applications in physics, economics, and artificial
intelligence, with advancements driven by computing.

Two-variable functions remain essential in mathematics and applied sciences.

8.1. Norm

Usual norm on ℝ𝟐 : Let 𝑥(𝑥1 , 𝑥2 ) 𝑎 𝑝𝑜𝑖𝑛𝑡 𝑖𝑛 the planeℝ𝟐 .

a) Manhattan norm: ‖(𝑥1 , 𝑥2 ) ‖1 = |𝑥1 | + |𝑥2 | .
b) Euclidean norm: ‖(𝑥1 , 𝑥2 ) ‖2 = √(𝑥1 )2 + (𝑥2 )2 .
c) Maximum norm: ‖(𝑥1 , 𝑥2 )‖∞ = 𝑚𝑎𝑥(|𝑥1 |, |𝑥2 |).

Equivalence of Norms. Two norms ‖𝑥‖ and‖𝑥‖′ are called equivalent if there exist two
positive scalars 𝑎 and b such that

If two norms are equivalent, then they induce the same notions of open sets (same topology). In
particular, if a sequence an converges to the limit a with respect to the norm, then this sequence
converges to the same limit with respect to the equivalent norm.

1
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

The three norms: ‖𝑣 ‖1 , ‖𝑣 ‖2 𝑎𝑛𝑑‖𝑣 ‖∞ are equivalent. This is a result of following

geometrical inequalities:

‖𝑣 ‖∞ ≤ ‖𝑣 ‖2 ≤ ‖𝑣 ‖1 , ‖𝑣 ‖2 ≤ √2‖𝑣 ‖∞ ,

‖𝑣 ‖1 ≤ 2‖𝑣 ‖∞ , ‖𝑣 ‖1 ≤ 2‖𝑣 ‖2 .

Distance
Let 𝐸 be a non-empty set. A distance (or metric) on 𝐸 is a function d: 𝐸 × 𝐸 → [0, +∞[ that satisfies
for all 𝑥, 𝑦 𝑎𝑛𝑑 𝑧 ∈ 𝐸, the following three properties:

Usual distance on ℝ𝟐 : Let 𝐴(𝑥𝐴 , 𝑦𝐴 ) and 𝐵(𝑥𝐵 , 𝑦𝐵 ) be two points in the planeℝ𝟐
a) Manhattan distance 𝑑1 (𝐴, 𝐵) = |𝑥𝐴 − 𝑥𝐵 | + |𝑦𝐴 − 𝑦𝐵 |.
b) Euclidean Distance: 𝑑2 (𝐴, 𝐵) = 𝑑((𝑥𝐴 , 𝑦𝐴 ), (𝑥𝐵 , 𝑦𝐵 )) = √(𝑥𝐴 − 𝑥𝐵 )2 + (𝑦𝐴 − 𝑦𝐵 )2 .
c) Maximum distance :𝑑∞ (𝐴, 𝐵) = 𝑚𝑎𝑥(|𝑥𝐴 − 𝑥𝐵 |, |𝑦𝐴 − 𝑦𝐵 |)

Theorem. Any norm induces a metric by 𝑑(𝑥, 𝑦) = ‖𝑥 − 𝑦‖.

𝑎) 𝑑1 (𝐴, 𝐵) = ‖(𝑥𝐴 − 𝑥𝐵 , 𝑦𝐴 − 𝑦𝐵 ) ‖1 = |𝑥𝐴 − 𝑥𝐵 | + |𝑦𝐴 − 𝑦𝐵 |

𝑏) 𝑑2 (𝐴, 𝐵) = ‖(𝑥𝐴 − 𝑥𝐵 , 𝑦𝐴 − 𝑦𝐵 ) ‖2 = √(𝑥𝐴 − 𝑥𝐵 )2 + (𝑦𝐴 − 𝑦𝐵 )2 .

𝑐) 𝑑∞ (𝐴, 𝐵) = ‖(𝑥𝐴 − 𝑥𝐵 , 𝑦𝐴 − 𝑦𝐵 ) ‖∞ = 𝑚𝑎𝑥(|𝑥𝐴 − 𝑥𝐵 |, |𝑦𝐴 − 𝑦𝐵 |)

Definition
Let A be a point in the plane ℝ2 , 𝑟 a strictly positive real number and 𝑑 a distance (or metric)
on ℝ2

 The open ball centered at A with radius 𝑟, denoted B(A, 𝑟), is defined as:
𝐵(𝐴, 𝑟) = { 𝑀 ∈ ℝ2 ∣ 𝑑(𝐴, 𝑀) < 𝑟 }.

 The closed ball centered at A with radius 𝑟, denoted 𝐵̅(A, 𝑟), is defined as:
𝐵̅ (𝐴, 𝑟) = {𝑀 ∈ ℝ2 ∣ 𝑑(𝐴, 𝑀) ≤ 𝑟}.

2
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

By symmetrizing we obtain the shapes below:

Open set in ℝ2 . We say that U is an open set of ℝ2 if, for any point A ∈ U, U contains an open
ball centered at A.

Examples:

3
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Bounded Sets
Definition

Definition of a Neighborhood in ℝ2

8.2. Functions of two real variables

A function of two real variables is a function that takes two real numbers as input and returns a real
number as output. It is generally written as
𝑓: ℝ2 ⟶ ℝ
(𝑥, 𝑦) ⟶ 𝑓(𝑥, 𝑦) = 𝑧
Example
𝑓: ℝ2 ⟶ ℝ
(𝑥, 𝑦) ⟶ 𝑓(𝑥, 𝑦) = 𝑒 𝑥 + 𝑦.

8.2.1. The domain, range, and graph of 𝑧 = 𝑓 (𝑥, 𝑦)

Definition1(domain of 𝑓). The set of points where the function is defined is called the
domain of the function 𝑓 .
Definition2( Range of the function). The range of the function is the set of its values 𝑓(𝑥, 𝑦)
for all (𝑥, 𝑦) in its domain. that is, {𝑓(𝑥, 𝑦)|(𝑥, 𝑦) ∈ 𝐷}.
Example1. Find the domain of the function defined by 𝑔(𝑥, 𝑦) = √4 − 𝑥 2 − 𝑦 2
Solution. The domain of g is
2 2
𝐷 = {(𝑥, 𝑦) ∈ ℝ |4 − 𝑥 2 − 𝑦 2 ≥ 0} = {(𝑥, 𝑦) ∈ ℝ |𝑥 2 + 𝑦 2 ≤ 4}
which is the disk with center (0, 0) and radius 2. (See the following Figure)

4
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Example2. 𝑓 (𝑥, 𝑦) = 𝑙𝑛(1 + 𝑥 + 𝑦)

𝐷 = {(𝑥, 𝑦) ∈ ℝ2 |1 + 𝑥 + 𝑦 > 0} = {(𝑥, 𝑦) ∈ ℝ2 |𝑦 > −𝑥 − 1}
To plot this set, we first plot the straight line with equation 1 + x + y = 0. We then determine
which side of the line is the set 1 + 𝑥 + 𝑦 > 0. Here it is above the line.

Definition3(Graph of function).
The graph of a function 𝑓 with two variables 𝑥 𝑎𝑛𝑑 𝑦 is the surface 𝑧 = 𝑓(𝑥, 𝑦) formed
by the points (𝑥, 𝑦, 𝑧) 𝑖𝑛 𝑥𝑦𝑧-space with (𝑥, 𝑦) in the domain of the function 𝑓 and
𝑧 = 𝑓(𝑥, 𝑦).

Level curves. The level curves of a function f of two variables are the curves with equations
𝑓(𝑥, 𝑦) = 𝑘, where k is a constant (in the range of 𝑓 ).

5
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

8.2.2. Limits of Functions of Two Variables

Let D is a region in the plane and f : D →R be a function. Suppose (a, b) ∈ ℝ2 is such that
B ((a, b), r) \ {(a, b)} ⊂ D for some r > 0.
The limit of f(x,y) as (x,y) approaches (a, b) is L if and only if the following ϵ−δ condition
holds.
For every ϵ > 0, there exists a δ > 0 such that for all (x, y) ∈ D with
0 <√(𝑥 − 𝑎)2 + (𝑦 − 𝑏)2 < δ ⇒|f(x, y)−L| < ϵ.
In this case, we write lim lima 𝑓(𝑥, 𝑦) = 𝐿 . We also say that L is the limit of
𝑓(𝑥, 𝑦) = 𝐿 𝑜𝑟 𝑥→
(𝑥,𝑦)→(𝑎,𝑏)
𝑦→𝑏
𝑓 𝑎𝑡 (𝑎, 𝑏).

Uniqueness of limit. If limit of 𝑓(𝑥, 𝑦) 𝑎𝑠 (𝑥, 𝑦) approaches (𝑎, 𝑏) exists, then it is unique.

Theorem. (Properties of Limits of Functions of two Variables)

We list these properties for functions of two variables. Similar properties hold for functions of
more variables. Let us assume that 𝑙1, 𝑙2 , and k are real numbers and
lim 𝑓(𝑥, 𝑦) = 𝑙1 and lim 𝑔(𝑥, 𝑦) = 𝑙2 ,, then the following hold.
(𝑥,𝑦)→(𝑎,𝑏) (𝑥,𝑦)→(𝑎,𝑏)

Example1. Find lim 𝑥 2 𝑦 + 𝑥𝑦

(𝑥,𝑦)→(2,1)
Combining the rules mentioned above allows us to do the following
lim 𝑥 2 𝑦 + 𝑥𝑦 = 22 × 1 + 2 × 1 = 6.
(𝑥,𝑦)→(2,1)

6
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

𝑥 2 −𝑥𝑦
Example2. Find lim
(𝑥,𝑦)→(0,0) √𝑥−√𝑦
0
It’s an indeterminate form of the form 0, we multiply by the conjugate we get:

Limit Along a Path

• For a function of one variable, there are only two directions for approaching a point; from
left and from right.
• Whereas for a function of two variables, there are infinitely many directions, and infinite
number of paths on which one can approach a point.
• If the limit exists, then 𝑓(𝑥, 𝑦) must approach the same limit no matter how (𝑥, 𝑦)
approaches (𝑎, 𝑏).
•Thus, if we can find two different paths of approach along which the function 𝑓(𝑥, 𝑦) has
different limits, then it follows that limit of 𝑓(𝑥, 𝑦) 𝑎𝑠 (𝑥, 𝑦) approaches (𝑎, 𝑏) does not
exist.

Possible paths through (a; b)

𝑥 2 −𝑦 2
Example1. Find lim
(𝑥,𝑦)→(0,0) 𝑥 2 +𝑦 2

1. Limit along the path 𝑥 = 0. Along this path, we have

𝑥2 − 𝑦2 −𝑦 2
lim = lim = −1.
(𝑥,𝑦)→(0,0) 𝑥 2 + 𝑦 2 (𝑥,𝑦)→(0,0) +𝑦 2
𝑥=0
2. Limit along the path 𝑦 = 0. Along this path, we have
𝑥2 − 𝑦2 𝑥2
lim = lim = 1.
(𝑥,𝑦)→(0,0) 𝑥 2 + 𝑦 2 (𝑥,𝑦)→(0,0) 𝑥 2
𝑥=0
𝑥 2 −𝑦 2
Conclusion: The limits are different, therefore lim does not exist.
(𝑥,𝑦)→(0,0) 𝑥 2 +𝑦 2

𝑥𝑦
Example2. Prove that lim does not exist.
(𝑥,𝑦)→(0,0) 𝑥 2 +𝑦 2

7
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

1. Limit along the path 𝑥 = 0. Along this path, we have

𝑥𝑦 0
lim 2 2
= lim = lim (0) = 0.
(𝑥,𝑦)→(0,0) 𝑥 + 𝑦 (𝑥,𝑦)→(0,0) 𝑦 2 (𝑥,𝑦)→(0,0)
𝑥=0
2. Limit along the path 𝑦 = 0. Along this path, we have
𝑥𝑦 0
lim = lim = lim (0) = = 0.
(𝑥,𝑦)→(0,0) 𝑥 2 + 𝑦 2 (𝑥,𝑦)→(0,0) 𝑥 2 (𝑥,𝑦)→(0,0)
𝑦=0
The both limits are 0. Next, we try along the path 𝑦 = 𝑥

𝑥𝑦
We obtained a different limit. So, lim does not exist.
(𝑥,𝑦)→(0,0) 𝑥 2 +𝑦 2
𝑥2𝑦
Example2. Prove that lim = 0 along the path 𝑦 = 𝑚𝑥 for all 𝑚 but this limit
(𝑥,𝑦)→(0,0) 𝑥 4 +𝑦 2
does not exist.

However, we will get a different answer along the path 𝑦 = 𝑥 2

𝑥2𝑦 𝑥4
lim = lim
(𝑥,𝑦)→(0,0) 𝑥 4 + 𝑦 2 (𝑥,𝑥 2 )→(0,0) 𝑥 4 + 𝑥 4
𝑦=𝑥 2
𝑥2𝑦 𝑥4 1
lim = lim = .
(𝑥,𝑦)→(0,0) 𝑥 4 + 𝑦 2 (𝑥,𝑥 2 )→(0,0) 2𝑥 4 2
𝑦=𝑥 2

𝑥2𝑦
This proves that lim does not exist.
(𝑥,𝑦)→(0,0) 𝑥 4 +𝑦 2
Remark. Make sure that the path you select goes through the point at which we are computing
the limit.

8
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Additional Techniques to Limits:

Polar coordinates

Proposition

𝑥 3 +𝑦 3
Example1. Using polar coordinates, Find lim
(𝑥,𝑦)→(0,0) 𝑥 2 +𝑦 2
𝑥 = 𝑟𝑐𝑜𝑠𝜃, 𝑦 = 𝑟𝑠𝑖𝑛𝜃

9
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Because |𝑐𝑜𝑠 3 𝜃 + 𝑠𝑖𝑛3 𝜃| ≤ |𝑐𝑜𝑠 3 𝜃| + |𝑠𝑖𝑛3 𝜃| ≤ 2

So, |𝑟(𝑐𝑜𝑠 3 𝜃 + 𝑠𝑖𝑛3 𝜃)| ≤ 2𝑟 → 0
Then
𝑥 3 +𝑦 3
lim = lim r(𝑐𝑜𝑠 3 𝜃 + 𝑠𝑖𝑛3 𝜃) = 0.
(𝑥,𝑦)→(0,0) 𝑥 2 +𝑦 2 𝑟→0+
𝑥𝑦
Example2. Using polar coordinates, Find lim
(𝑥,𝑦)→(0,0) 𝑥 2 +𝑦 2
𝑥 = 𝑟𝑐𝑜𝑠𝜃, 𝑦 = 𝑟𝑠𝑖𝑛𝜃.
𝑥𝑦 𝑟 2 𝑐𝑜𝑠𝜃𝑠𝑖𝑛𝜃
lim = lim+ r2 (𝑐𝑜𝑠2 𝜃+𝑠𝑖𝑛2 𝜃)= 𝑐𝑜𝑠𝜃𝑠𝑖𝑛𝜃.
(𝑥,𝑦)→(0,0) 𝑥 2 +𝑦 2 𝑟→0
𝜋 1
If 𝜃 = 0 𝑐𝑜𝑠𝜃𝑠𝑖𝑛𝜃 = 0, if 𝜃 = 𝑐𝑜𝑠𝜃𝑠𝑖𝑛𝜃 = 2, Since the limit depends on the angle,
4
𝑥𝑦
lim does not exist.
(𝑥,𝑦)→(0,0) 𝑥 2 +𝑦 2

Remark. if (𝑥, 𝑦) → (𝑎, 𝑏) the polar coordinates are as follow

𝑥 − 𝑎 = 𝑟𝑐𝑜𝑠𝜃
𝑦 − 𝑏 = 𝑟𝑠𝑖𝑛𝜃
(𝑦−1)2 𝑠𝑖𝑛𝑥
Example. lim .
(𝑥,𝑦)→(0,1) 𝑥 2 +(𝑦−1)2
Let 𝑥 = 𝑟𝑐𝑜𝑠𝜃 and 𝑦 − 1 = 𝑟𝑠𝑖𝑛𝜃.

(𝑦 − 1)2 𝑠𝑖𝑛𝑥 (𝑟𝑠𝑖𝑛𝜃)2 sin(𝑟𝑐𝑜𝑠𝜃)

lim = 𝑙𝑖𝑚
(𝑥,𝑦)→(0,1) 𝑥 2 + (𝑦 − 1)2 𝑟→0+ (𝑟𝑐𝑜𝑠𝜃)2 + (𝑟𝑠𝑖𝑛𝜃)2

(𝑠𝑖𝑛𝜃)2 sin(𝑟𝑐𝑜𝑠𝜃)
= 𝑙𝑖𝑚+
𝑟→0 (𝑐𝑜𝑠𝜃)2 + (𝑠𝑖𝑛𝜃)2

= 𝑙𝑖𝑚+(𝑠𝑖𝑛𝜃)2 sin(𝑟𝑐𝑜𝑠𝜃) = 0
𝑟→0
Because 𝜃 → 𝑠𝑖𝑛2 𝜃 is bounded and sin(𝑟𝑐𝑜𝑠𝜃) → 0 for all 𝜃

Squeeze Theorem
If 𝑔 (𝑥, 𝑦) ≤ 𝑓 (𝑥, 𝑦) ≤ ℎ (𝑥, 𝑦) for all (𝑥, 𝑦) ≠ (𝑎, 𝑏) in a disk centered at (𝑎, 𝑏)
and if lim 𝑔(𝑥, 𝑦) = lim ℎ(𝑥, 𝑦) = 𝑙 then lim 𝑓(𝑥, 𝑦) = 𝑙 .
(𝑥,𝑦)→(𝑎,𝑏) (𝑥,𝑦)→(𝑎,𝑏) (𝑥,𝑦)→(𝑎,𝑏)

10
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

There is also an equivalent of the squeeze theorem. we use it when we suspect that
lim 𝑓(𝑥, 𝑦) = 𝑙 .
(𝑥,𝑦)→(𝑎,𝑏)

Theorem. Suppose that |𝑓(𝑥, 𝑦) − 𝑙| ≤ 𝑔 (𝑥, 𝑦) for every (𝑥; 𝑦) inside a disk centered
at (𝑎; 𝑏) except maybe at (𝑎; 𝑏). If lim 𝑔(𝑥, 𝑦) = 0 then lim 𝑓(𝑥, 𝑦) = 𝑙 .
(𝑥,𝑦)→(𝑎,𝑏) (𝑥,𝑦)→(𝑎,𝑏)
Remark. we usually use certain key or basic inequalities such as:
1) |𝑥| ≤ √𝑥 2 + 𝑦 2 ;
2) |𝑦| ≤ √𝑥 2 + 𝑦 2 ;
𝑥2
3) ≤ 1;
𝑥 2 +𝑦 2
𝑥
4) ≤ 1;
𝑥+1
𝑥 2 +𝑦 2
5) |𝑥𝑦| ≤ .
2

(𝑦−1)2 𝑠𝑖𝑛𝑥
Example1. Find lim
(𝑥,𝑦)→(0,1) 𝑥 2 +(𝑦−1)2

(𝑦 − 1)2
| |≤1
𝑥 2 + (𝑦 − 1)2
Multiplying both sides by |𝑠𝑖𝑛𝑥|
(𝑦 − 1)2
| 2 | |𝑠𝑖𝑛𝑥| ≤ |𝑠𝑖𝑛𝑥|
𝑥 + (𝑦 − 1)2
(𝑦 − 1)2
| 2 𝑠𝑖𝑛𝑥| ≤ |𝑠𝑖𝑛𝑥|
𝑥 + (𝑦 − 1)2
(𝑦−1)2 𝑠𝑖𝑛𝑥
Since lim |𝑠𝑖𝑛𝑥|=0, it follows by the squeeze theorem that lim = 0.
(𝑥,𝑦)→(0,1) (𝑥,𝑦)→(0,1) 𝑥 2 +(𝑦−1)2

Iterated limits
Suppose that lim 𝑓(𝑥, 𝑦) = 𝑙 and that the following one-dimensional limits
(𝑥,𝑦)→(𝑎,𝑏)

Then

If

11
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Example

𝑥 2 −𝑦 2
Hence, the limit lim does not exist.
(𝑥,𝑦)→(0,0) 𝑥 2 +𝑦 2

8.2.3. Continuity
Definition1. A function 𝑓 is said to be continuous at a point (𝑎; 𝑏) if the following is true:
1. (𝑎; 𝑏) is in the domain of 𝑓.
2. lim 𝑓(𝑥, 𝑦) exist.
(𝑥,𝑦)→(𝑎,𝑏)
3. lim 𝑓(𝑥, 𝑦) = 𝑓(𝑎, 𝑏).
(𝑥,𝑦)→(𝑎,𝑏)

Example. Let 𝑓 be the function defined by

𝑓: ℝ2 ⟶ ℝ
𝑥𝑦 2
(𝑥, 𝑦) ≠ (0,0)
(𝑥, 𝑦) ⟶ 𝑓(𝑥, 𝑦) = {𝑥 2 + 𝑦 2
0 𝑖𝑓 (𝑥, 𝑦) = (0,0)
𝐷𝑓 = ℝ2 , we have
𝑦2
≤1
𝑥2 + 𝑦2
Multiplying by |𝑥|, we obtain

𝑥𝑦 2
| | ≤ |𝑥|
𝑥2 + 𝑦2
𝑥𝑦 2
Since lim |𝑥|=0, it follows by the squeeze theorem that lim = 0 = 𝑓(0,0)
(𝑥,𝑦)→(0,0) (𝑥,𝑦)→(0,0) 𝑥 2 +𝑦2
So, 𝑓 is said to be continuous at a point (0, 0).

Definition2 (discontinuous function). If a function 𝑓 is not continuous at a point (a; b), we say
that it is discontinuous at (a; b).

Definition3 (Continuity on a set). We say that a function 𝑓 is continuous on a set D if it is

continuous at every point in D.

12
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Theorem. The following results are true for multivariable functions:

𝑥𝑦 2
Example1. Where 𝑖𝑠 𝑡ℎ𝑒 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛 𝑓 𝑑𝑒𝑓𝑖𝑛𝑒𝑑 𝑏𝑦 𝑓 (𝑥, 𝑦) =𝑥 2 +𝑦 2 continuous?

Example2. Where is the function 𝑓 continuous?

8.2.4. Partial dérivatives

Définition

We can also write

𝜕𝑓 𝑓(𝑥, 𝑦0 ) − 𝑓(𝑥0 , 𝑦0 )
(𝑥0 , 𝑦0 ) = lim
𝜕𝑥 𝑥→𝑥0 𝑥 − 𝑥0
𝜕𝑓 𝑓(𝑥0 , 𝑦) − 𝑓(𝑥0 , 𝑦0 )
(𝑥0 , 𝑦0 ) = lim .
𝜕𝑦 𝑦→𝑦0 𝑦 − 𝑦0

13
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Exactly the same rules of differentiation apply as for a function of one variable. If we have a
𝜕𝑓
function of two variables 𝑓(𝑥; 𝑦) we treat y as a constant when calculating𝜕𝑥 , and treat 𝑥 as
𝜕𝑓
a constant when calculating 𝜕𝑦.
The first order partial derivatives are denoted by
𝜕𝑓 𝜕
= 𝜕𝑥 𝑓(𝑥, 𝑦) = 𝜕𝑥 𝑓(𝑥, 𝑦) = 𝑓𝑥 .
𝜕𝑥

𝜕𝑓 𝜕
= 𝑓(𝑥, 𝑦) = 𝜕𝑦 𝑓(𝑥, 𝑦) = 𝑓𝑦 .
𝜕𝑦 𝜕𝑦

Example1. Find the first order partial derivatives of the function 𝑓 defined on ℝ2 by
𝑓(𝑥, 𝑦) = 𝑦𝑠𝑖𝑛(𝑥𝑦)
2
Solution. For all (𝑥, 𝑦) ∈ ℝ , we have
𝜕𝑓 𝜕𝑓
= 𝑦 2 𝑐𝑜𝑠(𝑥𝑦), = 𝑠𝑖𝑛(𝑥𝑦) + 𝑥𝑦𝑐𝑜𝑠(𝑥𝑦).
𝜕𝑥 𝜕𝑦
Example2

𝜕𝑓 𝜕𝑓
Determine (0, 0) and (0, 0).
𝜕𝑥 𝜕𝑦
Solution
𝑓(𝑥, 𝑦0 ) − 𝑓(𝑥0 , 𝑦0 ) 𝑓(𝑥, 0) − 𝑓(0,0)
lim = lim
𝑥→𝑥0 𝑥 − 𝑥0 𝑥→0 𝑥−0
1
(𝑥 2 )𝑠𝑖𝑛 2
√𝑥 = lim 𝑥𝑠𝑖𝑛 1
= lim
𝑥→0 𝑥 𝑥→0 √𝑥 2
1 1
We have |𝑠𝑖𝑛 √𝑥 2 | ≤ 1 ⟹ |𝑥| |𝑠𝑖𝑛 √𝑥 2 | ≤ 1|𝑥|
1 𝜕𝑓
Since lim |𝑥| = 0 and by Squeeze theorem lim |𝑥| |𝑠𝑖𝑛 √𝑥 2 | = 0 then (0, 0) = 0.
𝑥→0 𝑥→0 𝜕𝑥
𝑓(𝑥0 , 𝑦) − 𝑓(𝑥0 , 𝑦0 ) 𝑓(0, 𝑦) − 𝑓(0,0)
lim = lim
𝑦→𝑦0 𝑦 − 𝑦0 𝑦→0 𝑦−0
1
(𝑦 2 )𝑠𝑖𝑛
√𝑦 2 1 𝜕𝑓
= lim = lim 𝑦𝑠𝑖𝑛 =0= (0, 0).
𝑥→0 𝑦 𝑥→0 √𝑦 2 𝜕𝑦

14
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Geometric interpretation of first order partial derivatives

𝜕𝑓
(𝑥0 , 𝑦0 )gives the slope of the tangent at the point(𝑥0 , 𝑦0 , 𝑓(𝑥0 , 𝑦0 ))to the curve of intersection
𝜕𝑥
the surface with a vertical plane y =𝑦0 .

𝜕𝑓
(𝑥0 , 𝑦0 )gives the slope of the tangent at the point(𝑥0 , 𝑦0 , 𝑓(𝑥0 , 𝑦0 ))to the curve of intersection
𝜕𝑦
Of the surface with a vertical plane 𝑥 = 𝑥0 .

Remark. The existence of partial derivatives at a point does not necessarily imply that the
function is continuous at that point.

Example. Let f be the function defined by

𝑓: ℝ2 ⟶ ℝ
𝑥𝑦
(𝑥, 𝑦) ≠ (0,0)
(𝑥, 𝑦) ⟶ 𝑓(𝑥, 𝑦) = { + 𝑦 2𝑥2
0 𝑖𝑓 (𝑥, 𝑦) = (0,0)

𝑓(𝑥, 0) − 𝑓(0,0) 0
lim = lim = 0
𝑥→0 𝑥−0 𝑥→0 𝑥

𝑓(0, 𝑦) − 𝑓(0,0) 0
lim = lim = 0.
𝑦→0 𝑦−0 𝑦→0 𝑦

So,
𝜕𝑓 𝜕𝑓
(0, 0) = (0, 0) = 0.
𝜕𝑥 𝜕𝑦

15
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

We will study the continuity of 𝑓 at (0,0) :

𝑥𝑦
lim = lim 0 = 0.
(𝑥,𝑦)→(0,0) 𝑥 2 + 𝑦 2 (𝑥,𝑦)→(0,0)
𝑥=0
2
𝑥 𝑥2 1 𝑥𝑦
lim = lim = ≠ lim
(𝑥,𝑦)→(0,0) 𝑥 2 + 𝑥 2 (𝑥,𝑦)→(0,0) 2𝑥 2 2 (𝑥,𝑦)→(0,0) 𝑥 2 + 𝑦 2
𝑥=𝑦 𝑥=0
𝑥𝑦
Therefore lim does not exist, then the function 𝑓 is not continuous at(0,0).
(𝑥,𝑦)→(0,0) 𝑥 2 +𝑦 2

8.2.5. C¹ Functions on a Subset 𝐷 of ℝ2

Definition
Let 𝐷 be a subset of ℝ2 and let 𝑓: ℝ2 ⟶ ℝ be a function defined on 𝐷. We say that 𝑓 is of
class 𝑪𝟏 on 𝐷 if
 It has first-order partial derivatives on 𝐷,
 And these two partial derivatives are continuous on 𝐷.

Example. let 𝑓: ℝ2 ⟶ ℝ be a function defined on 𝐷=]0, +∞[ × ]0, +∞[ by

𝑙𝑛𝑥
𝑓(𝑥, 𝑦) =
𝑦
𝜕𝑓 1 𝜕𝑓 −𝑙𝑛𝑥
For all (𝑥, 𝑦) ∈ ]0, +∞[ × ]0, +∞[: 𝜕𝑥 (𝑥, 𝑦) = 𝑥𝑦 𝜕𝑦 (𝑥, 𝑦) = 𝑦2
𝜕𝑓 𝜕𝑓
The function 𝑓 is of class C¹ on 𝐷 because 𝜕𝑥
and 𝜕𝑦
exist for all (𝑥, 𝑦) 𝑖𝑛 𝐷
𝑎𝑛𝑑 are continious as a product of continuous functions.
Theorem1
Let 𝐷 be a subset of ℝ2 and 𝑓: ℝ2 ⟶ ℝ be a function defined on 𝐷. If 𝑓 is of class 𝐶 1 on 𝐷,
then 𝑓 is continuous on 𝐷.

Theorem2
Every polynomial function is of class 𝑪𝟏 on ℝ2 .

Definition

16
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

𝑃 and 𝑄 are polynomial functions

Theorem3

Theorem4

17
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

8.2. 6. Higher Order Partial Derivatives

Example. Find all partials up to the second order of the function

𝑓(𝑥, 𝑦) = 𝑥 3 + 2𝑥 2 𝑦 + 𝑥𝑦 3 − 4𝑦 2 .
Solution
𝜕𝑓 𝜕𝑓 𝜕 2𝑓
(𝑥, 𝑦) = 3𝑥 2 + 4𝑥𝑦 + 𝑦 3 , (𝑥, 𝑦) = 2𝑥 2 + 3𝑥𝑦 2 − 8𝑦, (𝑥, 𝑦) = 6𝑥 + 4𝑦,
𝜕𝑥 𝜕𝑦 𝜕𝑥 2

18
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓
(𝑥, 𝑦) = 6𝑥𝑦 − 8𝑦, (𝑥, 𝑦) = 4𝑥 + 3𝑦 2 , (𝑥, 𝑦) = 4𝑥 + 3𝑦 2 .
𝜕𝑦 2 𝜕𝑥𝜕𝑦 𝜕𝑦𝜕𝑥

Exercise
Let 𝛼 ∈ ℝ and let 𝑓: ℝ2 ⟶ ℝ be defined by
𝑥2𝑦2
, (𝑥, 𝑦) ≠ (0,0);
𝑓(𝑥, 𝑦) = {(𝑥 2 + 𝑦 2 )𝛼
0, 𝑖𝑓 (𝑥, 𝑦) = (0,0).

1. Show that 𝑓 is continuous on ℝ2 if and only if 𝛼 < 2.

2. Show that for all 𝛼 ∈ ℝ, 𝑓 has partial derivativatives at every point in ℝ2 .
3. Determine all vallues of 𝛼 for which 𝑓is of class 𝐶 1 on ℝ2 .

3. For (𝑥, 𝑦) ≠ (0,0)

19
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

For (𝑥, 𝑦) = (0,0)

Therefore the first order partial derivatives are defined by

2𝑥𝑦 2 (𝑥 2 + 𝑦 2 − 𝛼𝑥 2 )
𝜕𝑓 , (𝑥, 𝑦) ≠ (0,0);
={ (𝑥 2 + 𝑦 2 )𝛼+1
𝜕𝑥
0, 𝑖𝑓 (𝑥, 𝑦) = (0,0).
2𝑦𝑥 2 (𝑥 2 + 𝑦 2 − 𝛼𝑦 2 )
𝜕𝑓 , (𝑥, 𝑦) ≠ (0,0);
={ (𝑥 2 + 𝑦 2 )𝛼+1
𝜕𝑦
0, 𝑖𝑓 (𝑥, 𝑦) = (0,0).
4.

We will study the continuity of the partial derivatives at (0,0), we have

𝜕𝑓 2𝑥𝑦 2 (𝑥 2 + 𝑦 2 − 𝛼𝑥 2 )
lim = lim
(𝑥,𝑦)⟶(0,0) 𝜕𝑥 𝑟⟶0 (𝑥 2 + 𝑦 2 )𝛼+1

𝑟 3 𝑐𝑜𝑠𝜃𝑠𝑖𝑛2 𝜃(𝑟 2 − 𝛼𝑟 2 𝑐𝑜𝑠 2 𝜃)

= lim
𝑟⟶0 r 2𝛼+2

𝑟 5 𝑐𝑜𝑠𝜃𝑠𝑖𝑛2 𝜃(1 − 𝛼𝑐𝑜𝑠 2 𝜃)

= lim
𝑟⟶0 r 2𝛼+2
𝜕𝑓
lim = lim 𝑟 3−2𝛼 𝑐𝑜𝑠𝜃𝑠𝑖𝑛2 𝜃(1 − 𝛼𝑐𝑜𝑠 2 𝜃).
(𝑥,𝑦)⟶(0,0) 𝜕𝑥 𝑟⟶0
And
𝜕𝑓 2𝑦𝑥 2 (𝑥 2 + 𝑦 2 − 𝛼𝑦 2 )
lim = lim
(𝑥,𝑦)⟶(0,0) 𝜕𝑦 𝑟⟶0 (𝑥 2 + 𝑦 2 )𝛼+1

𝜕𝑓
lim = lim 𝑟 3−2𝛼 𝑠𝑖𝑛𝜃𝑐𝑜𝑠 2 𝜃(1 − 𝛼𝑠𝑖𝑛2 𝜃).
(𝑥,𝑦)⟶(0,0) 𝜕𝑦 𝑟⟶0
We have
|𝑐𝑜𝑠𝜃𝑠𝑖𝑛2 𝜃(1 − 𝛼𝑐𝑜𝑠 2 𝜃)| ≤ 1 + |𝛼| and |𝑠𝑖𝑛𝜃𝑐𝑜𝑠 2 𝜃(1 − 𝛼𝑠𝑖𝑛2 𝜃)| ≤ 1 + |𝛼|.
So

𝑟 3−2𝛼 |𝑐𝑜𝑠𝜃𝑠𝑖𝑛2 𝜃(1 − 𝛼𝑠𝑖𝑛2 𝜃)| ≤ (1 + |𝛼|)𝑟 3−2𝛼

𝑟 3−2𝛼 |𝑠𝑖𝑛𝜃𝑐𝑜𝑠 2 𝜃(1 − 𝛼𝑐𝑜𝑠 2 𝜃)| ≤ (1 + |𝛼|)𝑟 3−2𝛼

3
• If 3 − 2𝛼 > 0 that is to say 𝛼 < 2 , lim (1 + |𝛼|)𝑟 3−2𝛼 = 0 then by squeeze theorem
𝑟⟶0

lim 𝑟 3−2𝛼 𝑐𝑜𝑠𝜃𝑠𝑖𝑛2 𝜃(1 − 𝛼𝑐𝑜𝑠 2 𝜃) = 0 and lim 𝑟 3−2𝛼 𝑠𝑖𝑛𝜃𝑐𝑜𝑠 2 𝜃(1 − 𝛼𝑠𝑖𝑛2 𝜃) = 0
𝑟⟶0 𝑟⟶0

20
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

So
𝜕𝑓 𝜕𝑓
lim (𝑥, 𝑦) = 0 = (0,0).
(𝑥,𝑦)⟶(0,0) 𝜕𝑥 𝜕𝑥
And
𝜕𝑓 𝜕𝑓
lim (𝑥, 𝑦) = 0 = (0,0).
(𝑥,𝑦)⟶(0,0) 𝜕𝑦 𝜕𝑦
𝜕𝑓 𝜕𝑓
In this case 𝜕𝑥 and 𝜕𝑦 are continious at (0,0).
3
• If 3 − 2𝛼 = 0 that is to say 𝛼 = 2
𝜕𝑓
lim = lim 𝑐𝑜𝑠𝜃𝑠𝑖𝑛2 𝜃(1 − 𝛼𝑐𝑜𝑠 2 𝜃)
(𝑥,𝑦)⟶(0,0) 𝜕𝑥 𝑟⟶0
𝜕𝑓 𝜋 𝜕𝑓 √2 𝜕𝑓
If 𝜃 = 0 lim = 0, If 𝜃 = 4 lim = 4 ≠ 0, so 𝜕𝑥 is not continuous at (0,0).
(𝑥,𝑦)⟶(0,0) 𝜕𝑥 𝜕𝑥
(𝑥,𝑦)⟶(0,0)

3
• If 3 − 2𝛼 < 0 that is to say 𝛼 > 2

3
𝜕𝑓 𝜕𝑓 2𝑥 5 (2−𝛼) 2𝑥 3−2𝛼 (2−𝛼) +∞ 𝑖𝑓 <𝛼<2
lim = lim (𝑥, 𝑥) = lim (2𝑥 2 )𝛼+1 = lim (2)𝛼+1
={ 2
(𝑥,𝑦)⟶(0,0) 𝜕𝑥 𝑥⟶0 𝜕𝑥 𝑥⟶0 𝑥⟶0 −∞ 𝑖𝑓 𝛼>2
𝑥=𝑦
𝜕𝑓
In this case is not continuous at (0,0).
𝜕𝑥
3
Conclusion: the function 𝑓 is of class 𝐶 1 on ℝ2 if and only if 𝛼 < 2.

8.2.7. Differentiability for functions of two variables

We begin by reviewing the concept of differentiation for functions of one variable.
Definition1. Let 𝑓: 𝐷 ⊂ ℝ ⟶ ℝ and let 𝑎 be an interior point of 𝐷. 𝑓 is differentiable at 𝑎
means
𝑓(𝑎 + ℎ) − 𝑓(𝑎)
lim ∈ℝ
ℎ⟶0 ℎ
′ (𝑎)
This limit is denoted by 𝑓 and this number is called the derivative of 𝑓 at 𝑎.

Equivalently

𝑓(𝑎 + ℎ) − 𝑓(𝑎) 𝑓(𝑥) − 𝑓(𝑎)

lim = lim = 𝑓 ′ (𝑎).
ℎ⟶0 ℎ 𝑥⟶𝑎 𝑥−𝑎

Geometrically the derivative of a function at 𝑎 is interpreted as the slope of the line tangent to
the graph of 𝑓 at the point (𝑎, 𝑓(𝑎)).

Extending the definition of differentiability in its present form to functions of several

variables is not possible because the definition involves division and dividing by a vector or
by a point in n dimensional space is not possible. To carry out the extension an equivalent
definition is developed that involves division by a distance. The limit statement can be
rewritten as

21
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

So, the following definition is equivalent to the original one.

Definition2. Let 𝑓: 𝐷 ⊂ ℝ ⟶ ℝ and let 𝑎 be an interior point of 𝐷. 𝑓 is differentiable at 𝑎

means there is a number 𝑓 ′ (𝑎), such that

Differentiability for functions of two variables

Definition1

Where lim 𝜀 (ℎ, 𝑘) = 0 and ∥⋅∥ is any norm on ℝ2 .

(ℎ,𝑘)→(0,0)

Equivalently

𝑓(𝑥0 +ℎ,𝑦0 +𝑘)−𝑓(𝑥0 ,𝑦0 )−ℎ𝐴−𝑘𝐵

lim =0 (If we use Euclidean norm)
(ℎ,𝑘)⟶(0,0) √ℎ2 +𝑘 2

The function

Called the differential of 𝑓 at the point (𝑥0 , 𝑦0 ).

Definition2. Let 𝑓: 𝐷 ⊂ ℝ2 ⟶ ℝ and let (𝑥0 , 𝑦0 ) be an interior point of 𝐷. 𝑓 is differentiable

at (𝑥0 , 𝑦0 )means there are two numbers 𝑓𝑥 (𝑎, 𝑏) 𝑎𝑛𝑑𝑓𝑦 (𝑎, 𝑏) such that
𝜕𝑓 𝜕𝑓
𝑓(𝑥0 + ℎ, 𝑦0 + 𝑘) − 𝑓(𝑥0 , 𝑦0 ) − ℎ (𝑥 , 𝑦 ) − 𝑘 (𝑥0 , 𝑦0 )
𝜕𝑥 0 0 𝜕𝑦
lim = 0.
(ℎ,𝑘)⟶(0,0) √ℎ2 + 𝑘 2

Or

22
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

𝜕𝑓 𝜕𝑓
𝑓(𝑥, 𝑦) − 𝑓(𝑥0 , 𝑦0 ) − (𝑥 − 𝑥0 ) (𝑥0 , 𝑦0 ) − (𝑦 − 𝑦0 ) (𝑥0 , 𝑦0 )
𝜕𝑥 𝜕𝑦
lim = 0.
(𝑥,𝑦)⟶(𝑎,𝑏) 2
√(𝑥 − 𝑥0 )2 + (𝑦 − 𝑦0 )

Geometric interpretation of differentiability at a point.

Interpret the definition of differentiability at a point (𝑥0 , 𝑦0 )as requiring that the graph of the
function 𝑓 have a tangent plane at the point (𝑥0 , 𝑦0 , 𝑓(𝑥0 , 𝑦0 )). The equation for this
tangent plane is

𝜕𝑓 𝜕𝑓
𝑧 = 𝑓(𝑥0 , 𝑦0 ) + (𝑥 − 𝑥0 ) (𝑥0 , 𝑦0 ) + (𝑦 − 𝑦0 ) (𝑥0 , 𝑦0 )
𝜕𝑥 𝜕𝑦

Example1

Since

23
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

And

Differentiability In an open set: We say that 𝑓 is differentiable on an open Ω if it is differentiable at

every point of Ω.

The application

It is called, in this case, the differential of 𝑓on Ω, and is denoted by 𝑑𝑓.

𝜕𝑓 𝜕𝑓
And is defined by 𝑑𝑓 = 𝜕𝑥 𝑑𝑥 + 𝑑𝑦.
𝜕𝑦

Example. Study the differentiability at every point of the function 𝑓 defined by

We need to study differentiability at (0,0):

We have

𝜕𝑓 𝜕𝑓
𝑓(𝑎 + ℎ, 𝑏 + 𝑘) − 𝑓(𝑎, 𝑏) − ℎ (𝑎, 𝑏) − 𝑘 (𝑎, 𝑏)
𝜕𝑥 𝜕𝑦
√ℎ2 + 𝑘 2

24
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

𝜕𝑓 𝜕𝑓
𝑓(0+ℎ,0+𝑘)−𝑓(0,0)−ℎ𝜕𝑥(0,0)−𝑘𝜕𝑦(0,0)
=
√ℎ2 +𝑘 2

𝑓(ℎ,𝑘)−𝑓(0,0)−ℎ+3𝑘
=
√ℎ2 +𝑘 2

ℎ4
ℎ−3𝑘+ 2 2 −ℎ+3𝑘
ℎ +𝑘
=
√ℎ2 +𝑘 2

ℎ4
= 3
(ℎ2 +𝑘 2 )2

We have

𝜕𝑓 𝜕𝑓
𝑓(0 + ℎ, 0 + 𝑘) − 𝑓(0,0) − ℎ (0,0) − 𝑘 (0,0) ℎ4
𝜕𝑥 𝜕𝑦
= 3
√ℎ2 + 𝑘 2 (ℎ2 + 𝑘 2 )2

And

ℎ4 ℎ4 ℎ4
0 ≤ 3 ≤ 3 = = |ℎ| →0
2 2 2 (|ℎ|)3 (ℎ,𝑘)⟶(0,0)
(ℎ + 𝑘 )2 (ℎ )2

So

𝜕𝑓 𝜕𝑓
𝑓(0+ℎ,0+𝑘)−𝑓(0,0)−ℎ (0,0)−𝑘 (0,0)
𝜕𝑥 𝜕𝑦
0≤ ≤ |ℎ| →0
√ℎ2 +𝑘 2 (ℎ,𝑘)⟶(0,0)

Then by squeeze theorem

𝜕𝑓 𝜕𝑓
𝑓(ℎ, 𝑘) − 𝑓(0,0) − ℎ (0,0) − 𝑘 (0,0)
𝜕𝑥 𝜕𝑦
lim = 0.
(ℎ,𝑘)⟶(0,0) √ℎ2 + 𝑘 2

Thus 𝑓 is differentiable at the point (0,0) and 𝑑𝑓(0,0) (ℎ, 𝑘) = ℎ − 3𝑘.

𝜕𝑓 𝜕𝑓
And the differential is defined by 𝑑𝑓 = 𝜕𝑥 𝑑𝑥 + 𝑑𝑦.
𝜕𝑦

Theorem. Let 𝑓 be a real-valued function defined on an open subset 𝐷 of ℝ2 . Let (𝑥0 , 𝑦0 ) ∈

𝐷. If 𝑓 is of class 𝐶 1 in the neighborhood of (𝑥0 , 𝑦0 ), then 𝑓 is differentiable at (𝑥0 , 𝑦0 ).
The converse is false.

⇒𝑓 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠
𝑓 ∈ 𝐶 1 ⟹ 𝑓 is differentiable 𝜕𝑓 𝜕𝑓
⇒ 𝑎𝑛𝑑 𝑒𝑥𝑖𝑠𝑡
𝜕𝑥 𝜕𝑦

25
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Next, for any (𝑥, 𝑦)∈ℝ2 :

8.2.8. Class 𝐶 2 functions of two variables on a domain

Definition

Theorem1

Theorem2

26
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Stability of the 𝐶 2 Property Under Algebraic Operations

Theorem.

Remark. It should be noted that the linear combination, the product, and the quotient
(provided that the denominator does not vanish) of 𝐶 2 functions on 𝐷 are still 𝐶 2 functions on
𝐷.

Stability of the 𝐶 2 Property Under Composition

Theorem

8.3. Hessian Matrix

Definition. Let 𝑓: ℝ × ℝ ⟶ ℝ be a function defined on 𝐷. We assume that 𝑓 has
Second-order partial derivatives. The Hessian matrix of 𝑓 at (𝑥, 𝑦) is the matrix denoted by
∇2 (𝑓)(𝑥, 𝑦), and defined as
𝜕 2𝑓 𝜕 2𝑓
𝜕𝑥 2 𝜕𝑥𝜕𝑦
∇2 (𝑓)(𝑥, 𝑦) = ( 2 )
𝜕 𝑓 𝜕 2𝑓
𝜕𝑦𝜕𝑥 𝜕𝑦 2
2
Example. let 𝑓: ℝ ⟶ ℝ be defined by
𝑓(𝑥, 𝑦) = 𝑥 3 + 2𝑥 2 𝑦 + 𝑥𝑦 3 − 4𝑦 2 .

27
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Solution
Step 1: First-order partial derivatives
𝜕𝑓 𝜕𝑓
(𝑥, 𝑦) = 3𝑥 2 + 4𝑥𝑦 + 𝑦 3 , (𝑥, 𝑦) = 2𝑥 2 + 3𝑥𝑦 2 − 8𝑦,
𝜕𝑥 𝜕𝑦
Step 2: Second-order partial derivatives
𝜕2 𝑓 𝜕2 𝑓 𝜕2 𝑓 𝜕2 𝑓
(𝑥, 𝑦) = 6𝑥 + 4𝑦, (𝑥, 𝑦) = 6𝑥𝑦 − 8𝑦, (𝑥, 𝑦) = (𝑥, 𝑦) = 4𝑥 + 3𝑦 2 .
𝜕𝑥 2 𝜕𝑦 2 𝜕𝑥𝜕𝑦 𝜕𝑦𝜕𝑥

Step 3: Hessian matrix

2 (𝑓)(𝑥,
6𝑥 + 4𝑦 4𝑥 + 3𝑦 2
∇ 𝑦) = 𝐻(𝑓)(𝑥, 𝑦) = ( )
4𝑥 + 3𝑦 2 6𝑥𝑦 − 8𝑦

Schwarz's Theorem (Symmetry of Mixed Partial Derivatives)

8.4. Derivatives of Composite Functions – Chain Rule.

Definition

is differentiable, and the chain rule states that the derivative of the composite function
𝑔(𝑡) = 𝑓(𝑥(𝑡), 𝑦(𝑡))is given by:

28
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

In General
A tree diagram is a visual way to organize information that may help you remember “which
derivatives go where” in the Chain Rule formula. Write the name of the first function at the
top of the diagram. Draw branches to x and y, the names of the independent variables of 𝑓.
Then draw more branches to t, the independent variable of 𝑥 and 𝑦. Then add in the
derivative notations as shown in the diagram below.

Example. Calculate 𝑔′(𝑡) such that

𝑓(𝑥, 𝑦) = 𝑥 2 + 2𝑥𝑦, 𝑔(𝑡) = 𝑓(𝑥(𝑡), 𝑦(𝑡)), 𝑥(𝑡) = 𝑡 2 , 𝑦(𝑡) = 𝑒 𝑡 .

Solution: we have
𝜕𝑓 𝜕𝑓 𝜕𝑥 𝜕𝑦
= 2𝑥 + 2𝑦, = 2𝑥, = 2𝑡, = 𝑒𝑡.
𝜕𝑥 𝜕𝑦 𝜕𝑡 𝜕𝑡

𝑔′ (𝑡) = (2𝑥(𝑡) + 2𝑦(𝑡))2𝑡 + (2𝑥(𝑡))𝑒 𝑡

𝑔′ (𝑡) = (2𝑡 2 + 2𝑒 𝑡 )2𝑡 + (2𝑡 2 )𝑒 𝑡 .

We can define the composite function:

When the first-order partial derivatives involved are defined, we have:

29
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Example.

Solution:

Exercise. 𝐿𝑒𝑡 𝑧 = 𝑓 (𝑥, 𝑦) = 𝑥 2 𝑦 𝑤ℎ𝑒𝑟𝑒 𝑥 = 𝑟 𝑐𝑜𝑠(𝜃) 𝑎𝑛𝑑 𝑦 = 𝑟 𝑠𝑖𝑛(𝜃).

𝜕𝑧 𝜕𝑧
Compute 𝜕𝑟 and 𝜕𝜃.
Solution
𝜕𝑓 𝜕𝑓 𝜕𝑥 𝜕𝑦
= 2𝑥𝑦, =𝑥 2 , 𝜕𝑟 = 𝑐𝑜𝑠(𝜃), = 𝑠𝑖𝑛(𝜃)
𝜕𝑥 𝜕𝑦 𝜕𝑟
𝜕𝑧
We compute 𝜕𝑟

𝜕𝑧 𝜕𝑓 𝜕𝑥 𝜕𝑓 𝜕𝑦
= + = 2𝑥𝑦𝑐𝑜𝑠𝜃 + 𝑥 2 𝑠𝑖𝑛𝜃.
𝜕𝑟 𝜕𝑥 𝜕𝑟 𝜕𝑦 𝜕𝑟
𝜕𝑧
= 2𝑟 𝑐𝑜𝑠(𝜃) 𝑟 𝑠𝑖𝑛(𝜃) 𝑐𝑜𝑠(𝜃) + 𝑟 2 𝑐𝑜𝑠 2 (𝜃)𝑠𝑖𝑛𝜃.
𝜕𝑟
𝜕𝑧
= 3𝑟 2 𝑐𝑜𝑠 2 (𝜃) 𝑠𝑖𝑛(𝜃)
𝜕𝑟
𝜕𝑧
Now we compute 𝜕𝜃

𝜕𝑧 𝜕𝑓 𝜕𝑥 𝜕𝑓 𝜕𝑦
= +
𝜕𝜃 𝜕𝑥 𝜕𝜃 𝜕𝑦 𝜕𝜃

30
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

𝜕𝑧
= 2𝑥𝑦(−𝑟𝑠𝑖𝑛𝜃) + 𝑥 2 𝑟 𝑐𝑜𝑠(𝜃)
𝜕𝜃
𝜕𝑧
= 2𝑟 𝑐𝑜𝑠(𝜃) 𝑟 𝑠𝑖𝑛(𝜃)(−𝑟𝑠𝑖𝑛𝜃) + 𝑟 2 𝑐𝑜𝑠 2 (𝜃)𝑟 𝑐𝑜𝑠(𝜃)
𝜕𝜃
𝜕𝑧
= −2𝑟 3 𝑐𝑜𝑠(𝜃) 𝑠𝑖𝑛2 (𝜃) + 𝑟 3 𝑐𝑜𝑠 3 (𝜃).
𝜕𝜃
Method 2
𝑧 = 𝑓 (𝑥, 𝑦) = 𝑥 2 𝑦 = 𝑟 3 𝑐𝑜𝑠 2 (𝜃)𝑠𝑖𝑛𝜃.
𝜕𝑧
= 3𝑟 2 𝑐𝑜𝑠 2 (𝜃) 𝑠𝑖𝑛(𝜃)
𝜕𝑟
𝜕𝑧
= 𝑟 3 (−2 𝑐𝑜𝑠(𝜃) 𝑠𝑖𝑛2 (𝜃) + 𝑐𝑜𝑠 3 (𝜃)).
𝜕𝜃

8.5. Jacobian matrix

8.5.1. Definition

31
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Particular case

8.5.2. Jacobian of Composite function

Let us first recall the chain rule formula for functions from ℝ to ℝ.

Proposition

Theorem

Example. Let 𝑓: ℝ2 → ℝ2 , defined by 𝑓(𝑥, 𝑦) = (𝑥 + 2𝑦, 𝑒 𝑥−𝑦 ) and 𝑔: ℝ2 → ℝ, defined by

𝑔(𝑥, 𝑦) = 𝑥𝑐𝑜𝑠𝑦
1) Determine the Jacobian matrix of 𝑓 and the Jacobian matrix of 𝑔.
2) Determine the Jacobian matrix of 𝑔𝑜𝑓 by two methods.
Solution
𝜕𝑓1 𝜕𝑓1
𝜕𝑥 𝜕𝑦
1) The Jacobian matrix of 𝑓 is 𝐽𝑓 (𝑥, 𝑦) = ( 𝜕𝑓2 𝜕𝑓2
) such that
𝜕𝑥 𝜕𝑦
𝑓1 (𝑥, 𝑦) = 𝑥 + 2𝑦 and 𝑓2 (𝑥, 𝑦) = 𝑒 𝑥−𝑦

32
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

So
𝐽𝑓 (𝑥, 𝑦) = (𝑒 𝑥−𝑦1 2
−𝑒 𝑥−𝑦
).

𝜕𝑔 𝜕𝑔
The Jacobian matrix of 𝑔 is 𝐽𝑔 (𝑥, 𝑦) = (𝜕𝑥 ) = (𝑐𝑜𝑠𝑦 − 𝑥𝑠𝑖𝑛𝑦).
𝜕𝑦
2) To obtain the Jacobian matrix of the composition 𝑔𝑜𝑓: ℝ2 → ℝ , we apply the formula
for the product of matrix:

𝐽𝑔𝑜𝑓 (𝑥, 𝑦) = 𝐽𝑔 (𝑓(𝑥, 𝑦)) × 𝐽𝑓 (𝑥, 𝑦)

𝐽𝑔𝑜𝑓 (𝑥, 𝑦) = 𝐽𝑔 ((𝑥 + 2𝑦 , 𝑒 𝑥−𝑦 )) × 𝐽𝑓 (𝑥, 𝑦)

1 2
𝐽𝑔𝑜𝑓 (𝑥, 𝑦) = (𝑐𝑜𝑠(𝑒 𝑥−𝑦 ) − (𝑥 + 2𝑦)𝑠𝑖𝑛(𝑒 𝑥−𝑦 )) ( )
𝑒 𝑥−𝑦 − 𝑒 𝑥−𝑦

𝐽𝑔𝑜𝑓 (𝑥, 𝑦) = (𝑐𝑜𝑠(𝑒 𝑥−𝑦 ) − 𝑒 𝑥−𝑦 (𝑥 + 2𝑦)𝑠𝑖𝑛(𝑒 𝑥−𝑦 ) 2𝑐𝑜𝑠(𝑒 𝑥−𝑦 ) + 𝑒 𝑥−𝑦 (𝑥 + 2𝑦)𝑠𝑖𝑛(𝑒 𝑥−𝑦 ))

Second method
𝑔𝑜𝑓(𝑥, 𝑦) = 𝑔(𝑓(𝑥, 𝑦))

= 𝑔(𝑥 + 2𝑦 , 𝑒 𝑥−𝑦 )

= (𝑥 + 2𝑦)𝑐𝑜𝑠( 𝑒 𝑥−𝑦 ).

𝜕𝑔𝑜𝑓
= 𝑐𝑜𝑠( 𝑒 𝑥−𝑦 ) − 𝑒 𝑥−𝑦 𝑠𝑖𝑛( 𝑒 𝑥−𝑦 )(𝑥 + 2𝑦),
𝜕𝑥
𝜕𝑔𝑜𝑓
= 2𝑐𝑜𝑠( 𝑒 𝑥−𝑦 ) + 𝑒 𝑥−𝑦 𝑠𝑖𝑛( 𝑒 𝑥−𝑦 )(𝑥 + 2𝑦),
𝜕𝑦

We have 𝑔𝑜𝑓: ℝ2 → ℝ , so the Jacobian matrix of the composition 𝑔𝑜𝑓 has one row and two
columns.
𝜕𝑔𝑜𝑓 𝜕𝑔𝑜𝑓
𝐽𝑔𝑜𝑓 (𝑥, 𝑦) = ( ))
𝜕𝑥 𝜕𝑦

𝐽𝑔𝑜𝑓 (𝑥, 𝑦) = (𝑐𝑜𝑠(𝑒 𝑥−𝑦 ) − 𝑒 𝑥−𝑦 (𝑥 + 2𝑦)𝑠𝑖𝑛(𝑒 𝑥−𝑦 ) 2𝑐𝑜𝑠(𝑒 𝑥−𝑦 ) + 𝑒 𝑥−𝑦 (𝑥 + 2𝑦)𝑠𝑖𝑛(𝑒 𝑥−𝑦 )).

8.6. Taylor formula (in two variables)

8.6.1. First-order Taylor Formula

We have already seen that one way to express that 𝑓 is differentiable is to say that 𝑓 admits a
first-order Taylor expansion.

33
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

8.6.2. Second-order Taylor formula (in two variables)

Theorem. Let 𝑓: 𝑈 → ℝ be a 𝐶 2 function on an open set 𝑈 ⊂ ℝ2 , and let (𝑥0 , 𝑦0 ) ∈ 𝑈. Then:

Another form of Taylor’s Theorem

If we put 𝑥 = 𝑥0 + ℎ, 𝑦 = 𝑦0 + 𝑘 then ℎ = 𝑥 − 𝑥0 , 𝑘 = 𝑦 − 𝑦0

We obtain

𝜕𝑓 𝜕𝑓
𝑓(𝑥, 𝑦) = 𝑓(𝑥0 , 𝑦0 ) + (𝑥 − 𝑥0 ) (𝑥0 , 𝑦0 ) + (𝑦 − 𝑦0 ) (𝑥0 , 𝑦0 )
𝜕𝑥 𝜕𝑥
1 𝜕2 𝑓 𝜕2 𝑓 𝜕2 𝑓
+2 ቂ(𝑥 − 𝑥0 )2 𝜕𝑥 2 (𝑥0 , 𝑦0 ) + 2(𝑥 − 𝑥0 )(𝑦 − 𝑦0 ) 𝜕𝑥𝜕𝑦 (𝑥0 , 𝑦0 ) + (𝑦 − 𝑦0 )2 𝜕𝑦 2 (𝑥0 , 𝑦0 )ቃ

+o(‖(𝑥 − 𝑥0 , 𝑦 − 𝑦0 )‖2 ).

• We also say that 𝑓 admits a second-order Taylor expansion at the point (𝑥0 , 𝑦0 )

• Recall that if we choose the Euclidean norm, then ∥ (ℎ, 𝑘) ∥2 =ℎ2 + 𝑘 2, and that 𝑜(∥ (ℎ, 𝑘) ∥2 )
refers to a function equal to ∥ (ℎ, 𝑘) ∥2 𝜀(ℎ, 𝑘) with 𝜀(ℎ, 𝑘) → 0 as (ℎ, 𝑘) → (0,0).

34
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Example1

Example2
Compute a first and second-order Taylor’s expansion around point (𝑎, 𝑏) = (1, −1) of a
function 𝑓 and give an approximate value of 𝑓(1.01, −0.98).

𝑓(𝑥, 𝑦) = 𝑒 𝑥𝑦

35
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Firstly, we compute all partial derivatives we need

𝑓(1, −1) = 𝑒 −1
𝜕𝑓 𝜕𝑓
(𝑥, 𝑦) = 𝑦𝑒 𝑥𝑦 , (𝑥, 𝑦) = 𝑥𝑒 𝑥𝑦
𝜕𝑥 𝜕𝑦

𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓 𝜕 2𝑓
(𝑥, 𝑦) = 𝑦 2 𝑒 𝑥𝑦 , (𝑥, 𝑦) = 𝑥𝑦𝑒 𝑥𝑦 = (𝑥, 𝑦), (𝑥, 𝑦) = 𝑥 2 𝑒 𝑥𝑦
𝜕𝑥 2 𝜕𝑥𝜕𝑦 𝜕𝑥𝜕𝑦 𝜕𝑦 2

So
𝜕𝑓 𝜕𝑓
(1, −1) = −𝑒 −1 , (1, −1) = 𝑒 −1
𝜕𝑥 𝜕𝑦

𝜕2 𝑓 𝜕2 𝑓 𝜕2 𝑓 𝜕2 𝑓
2
(1, −1) = 𝑒 −1 , (1, −1) = −𝑒 −1 = (1, −1), (1, −1) = 𝑒 −1 .
𝜕𝑥 𝜕𝑥𝜕𝑦 𝜕𝑥𝜕𝑦 𝜕𝑦 2

First -order Taylor expansion

𝜕𝑓 𝜕𝑓
𝑓(𝑥0 + ℎ, 𝑦0 + 𝑘) = 𝑓(𝑥0 , 𝑦0 ) + ℎ (𝑥0 , 𝑦0 ) + 𝑘 (𝑥0 , 𝑦0 ) + 𝑜(‖(ℎ, 𝑘)‖)
𝜕𝑥 𝜕𝑥
𝜕𝑓 𝜕𝑓
𝑓(1 + ℎ, −1 + 𝑘) = 𝑓(1, −1) + ℎ 𝜕𝑥 (1, −1) + 𝑘 𝜕𝑦 (1, −1) + 𝑜(‖(ℎ, 𝑘)‖).

𝑓(1 + ℎ, −1 + 𝑘) = 𝑒 −1 − ℎ𝑒 −1 − 𝑘𝑒 −1 + 𝑜(‖(ℎ, 𝑘)‖).

An approximate value of 𝑓(1.01, −0.98).

We have
ℎ = 𝑥 − 𝑥0 , 𝑘 = 𝑦 − 𝑦0

ℎ = 1,01 − 1 = 0,01, 𝑘 = −0.98 − (−1) = 0,02.

So
𝑓(1.01, −0.98) = 𝑓(1 + 0,01, −1 + 0,02) ≈ 𝑒−1 − 0,01𝑒−1 − 0,02𝑒−1 = 0,97𝑒−1 .

Second-order Taylor expansion

1
𝑓(1 + ℎ, −1 + 𝑘) = 𝑒 −1 − ℎ𝑒 −1 − 𝑘𝑒 −1 +2 [ℎ2 𝑒−1 − 2ℎ𝑘𝑒−1 + 𝑘 2 𝑒−1 ] + o(‖(ℎ, 𝑘)‖2 ).
An approximate value of 𝑓(1.01, −0.98):

𝑓(1.01, −0.98) = 𝑓(1 + 0,01, −1 + 0,02)

1
≈ 𝑒−1 − 0,01𝑒−1 − 0,02𝑒−1 + [0,012 𝑒−1 − 2 × 0,01 × 0,02𝑒−1 + 0,022 𝑒−1 ].
2

36
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

8.6.3. Taylor expansion of a function of two variables to order 𝑛

Example. Compute a third-order Taylor’s expansion around point (𝑎, 𝑏) = (0,0) of a

function 𝑓 defined by

Solution

37
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

1 𝑦3
𝑓(𝑥, 𝑦) = 0 + 𝑥(0) + 𝑦(1) + 2𝑥𝑦(1) + 2 𝑥 2 𝑦 − 6 + o(‖(ℎ, 𝑘)‖3 )
1 𝑦3
𝑓(𝑥, 𝑦) = 0 + 2𝑥𝑦 + 𝑥 2 𝑦 − + o(‖(ℎ, 𝑘)‖3 ).
2 6

8.6.4. Critical point (function of two variables)

Definition

Example
Determine the critical points of the function

Solution
1. Compute the partial derivatives:

38
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

So, the only critical point is (0,0)

8.6.5. Regular point. A point (𝑥0 , 𝑦0 ) is regular for the function 𝑓 if at least one of the
partial derivatives is nonzero at this point.

8.6.6. Minimum and Maximum: Case of Two Variables

Definition

Proposition

Theorem

• If 𝑟𝑡 − 𝑠 2 > 0 𝑎𝑛𝑑 𝑟 > 0, 𝑓(𝑥0 , 𝑦0 ) is local minimum.

• If 𝑟𝑡 − 𝑠 2 > 0 𝑎𝑛𝑑 𝑟 < 0, 𝑓(𝑥0 , 𝑦0 ) is local maximum.

• If 𝑟𝑡 − 𝑠 2 < 0 , 𝑓(𝑥0 , 𝑦0 ) is neither local minimum nor local maximum the point (𝑥0 , 𝑦0 ) is a
saddle point.
• If 𝑟𝑡 − 𝑠 2 = 0, no definitive conclusion can be drawn based on second-order conditions alone.

39
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

Proof

If 𝑟 ≠ 0, we can write

𝑠𝑘 2 𝑟𝑡 − 𝑠 2 2
𝑞(ℎ, 𝑘) = 𝑟 ((ℎ + ) + 𝑘 )
𝑟 𝑟2
From this, we conclude the following:
• If 𝑟𝑡 − 𝑠 2 > 0 𝑎𝑛𝑑 𝑟 > 0, then 𝑞(ℎ, 𝑘) > 0 for all vectors (ℎ, 𝑘) ≠ (0,0).
Therefore, 𝑓(𝑥0 , 𝑦0 ) is local minimum.

• If 𝑟𝑡 − 𝑠 2 > 0 𝑎𝑛𝑑 𝑟 < 0, then 𝑞(ℎ, 𝑘) > 0 for all vectors (ℎ, 𝑘) ≠ (0,0).
Therefore, 𝑓(𝑥0 , 𝑦0 ) is local minimum

40
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

• If 𝑟𝑡 − 𝑠 2 < 0 𝑎𝑛𝑑 , the sigh of 𝑞(ℎ, 𝑘) varies dependly on the relative values of ℎ 𝑎𝑛𝑑 𝑘. In this
case the point (𝑥0 , 𝑦0 ) is a saddle point.

• If 𝑟𝑡 − 𝑠 2 = 0, no definitive conclusion can be drawn based on second-order conditions alone.

Example1 :

Its first- and second-order partial derivatives are:

There exist four critical points are: 𝐴(1,2), 𝐵(−1, −2), 𝐶(2,1), 𝐷(−2, −1).

𝜕2 𝑓 𝜕2 𝑓 𝜕2 𝑓
For 𝐴(1,2) ቂ𝑟 = 𝜕𝑥 2 (1,2) = 6 , 𝑠 = 𝜕𝑥𝜕𝑦 = 12 , 𝑡 = 𝜕𝑦 2 (1,2) = 6ቃ.
𝜕2 𝑓 𝜕2 𝑓 𝜕2 𝑓
For 𝐵(−1, −2) ቂ𝑟 = 𝜕𝑥 2 (1,2) = −6 , 𝑠 = 𝜕𝑥𝜕𝑦 = −12 , 𝑡 = 𝜕𝑦 2 (1,2) = −6ቃ.
For points A and B: 𝑟𝑡 – 𝑠 2 = −108 < 0 therefore, f does not have an extremum at these
points.
𝜕2 𝑓 𝜕2 𝑓 𝜕2 𝑓
For 𝐶(2,1) ቂ𝑟 = 2
(1,2) = 12 , 𝑠 = = 6 ,𝑡 = (1,2) = 12ቃ
𝜕𝑥 𝜕𝑥𝜕𝑦 𝜕𝑦 2

𝑟𝑡 – 𝑠² = 108 > 0 𝑎𝑛𝑑 𝑟 > 0: the function f has a local minimum at the point (2, 1), and
this minimum is f(2, 1) = –28.

𝜕2 𝑓 𝜕2 𝑓 𝜕2 𝑓
For 𝐷(−2, −1) ቂ𝑟 = 𝜕𝑥 2 (1,2) = −12 , 𝑠 = 𝜕𝑥𝜕𝑦 = −6 , 𝑡 = 𝜕𝑦 2 (1,2) = −12ቃ

𝑟𝑡 – 𝑠² = 108 > 0 𝑎𝑛𝑑 𝑟 < 0: the function f has a local maximum at the point (–2, –1),
and this maximum is 𝑓(−2, −1) = 28.

41
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

42
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

43
National Polytechnic School
Preparatory cycle-ENP Math Analysis2 F.FELLAG

44