Calculus on Rn

Seminar 4 2020

Limits and continuity

Definition: Let A ⊆ Rn and f : A → Rm be a function. Let a ∈ A0
and b ∈ Rm . The point b is said to be the limit of the function f
at a, if

∀ε > 0, ∃δ > 0 s.t. ∀x ∈ A\{a}, with kx−ak < δ, it holds kf (x)−f (a)k < ε.
The usual notation for this is

lim f (x) = b.
x→a

During the first semester, we used a sequential definition, and we gave the ε as a
characterization theorem. This term is the other way arround. Now the definition is with
ε, and in practice, especially when we try to prove that the limit does not exist, we use
the following sequential characterization of limits:

lim f (x) = b ⇐⇒ ∀(xk ) ⊆ A\{a}, with lim xk = a it holds lim f (xk ) = b.

x→b k→∞ k→∞

Obviously, in order for us to prove that the limit at a certain

a ∈ A does not exist, we may emphasize two sequences (ak ) and
(bk ) from A\{a} such that

lim ak = lim bk = a but lim f (ak ) 6= lim f (bk ).

k→∞ k→∞ k→∞ k→∞

Recall that during the first semester, we used for the common
limit n → ∞. This term we usually have k → ∞. This is due to
the fact that we work in Rn , so n is now just a fixed number.

Exercise 1: Compute
sin(xy)
lim .
(x,y)→(0,2) x

Solution: We use for the solution, the sequential characterization.

1
 Let (xk , yk ) ⊆ R2 \{0, 2)} be a sequence such that

lim (xk , yk ) = (0, 2).

k→∞

This is equivalent to the fact that limk→∞ xk = 0 and limk→∞ yk = 2. Then

sin(xk yk ) sin(xk yk )
lim = lim · yk = 1 · lim yk = 1 · 2 = 2.
k→∞ xk k→∞ xk y k k→∞

sin(xk yk )
This is due to the fact that lim xk yk = 0 · 2 = 0, and obviously lim = 1. Since
k→∞ k→∞ xk yk
the sequence (xk , yk ) was chosen randomly, the proof is complete.

Excercise 2: Compute the following limits

x2 + y 2
a) lim p
(x,y)→(0,0) 1 + x2 + y 2 − 1
1
−
e x2 +y2
b) lim
(x,y)→(0,0) x4 + y 4

Solution
We treat both a) and b) with (x, y) → (0, 0) directly.

p
x2 + y 2 (x2 + y 2 )( 1 + x2 + y 2 + 1)
a) lim p = lim =
(x,y)→(0,0) 1 + x2 + y 2 − 1 (x,y)→(0,0) x2 + y 2
p
lim 1 + x2 + y 2 + 1 = 2.
(x,y)→(0,0)

b) Let us recall the classical mean inequality. Given a1 , ..., an > 0 then
r
n √ a 1 + ... + a n a21 + ... + a2n
1 ≤ n
a a
1 2 ...an ≤ ≤ .
a1
+ ... + a1n n n

We apply the last for x2 and y 2 . Thus

r
x2 + y 2 x4 + y 4
≤ .
2 2
We put this to the second power and exchange factors. Thus we get
1 2
≤ .
x4 + y 4 (x2 + y 2 )2

2
 1
−
If we multiply this by e x2 +y 2 , and get
1 1
− −
e x2 +y2 2e x2 +y2
≤ .
x4 + y 4 (x2 + y 2 )2

Due to the fact that (x, y) → (0, 0) we may substitute

1
to t, and t → ∞.
x2 + y2
So 1
−
2e x2 +y2 2e−t 2t2
= lim = lim = 0.
(x2 + y 2 )2 t→∞ t12 t→∞ et

Excercise 3: Study
xy
lim .
(x,y)→(0,0) x2 + y 2

Solution:

This is actually an exercise where we show that the limit does not exist.
The idea behind solving such exercising is passing the problem stated in two variables
(x, y) to a problem in just one variable. Since (x, y) → (0, 0) we try to consider some
particular instances.
The solution relies on emphasising a sequence (xk , yk ) → (0, 0) for which limk→∞ f (xk , yk )
does not exists.
Consider a random sequence (xk ) ⊂ R such that limk→∞ xk = 0.. Moreover, take
m ∈ R \{0}, and consider

yk = m · x k , ∀k ∈ N. Thus lim yk = 0,
k→∞

and therefore
lim (xk , yk ) = (0, 0).
k→∞

We get
xk · mxk m
lim f (xk , yk ) = lim 2 2 2
= .
k→∞ k→∞ x + m x
k k 1 + m2
We may assign to m different, values, and get for m = 1 and for m = 2
1 2
lim f (xk , 1 · xk ) = 6= = lim f (xk , 2 · xk ).
k→∞ 2 5 k→∞
Thus the limit does not exists.
Excercise 4: Study
x3 + y 3
lim .
(x,y)→(0,0) x2 + y 2

3
 Solution:

Let (xk , yk ) ∈ R2 \{(0, 0)} be s.t.

lim (xk , yk ) = (0, 0).

k→∞

We notice that
 3
 xk + yk3  3 3
 3
 xk + yk3  2 2
 
|x k | |y k |  ≤ |xk | xk +|yk | yk
0 ≤  2 ≤ + ≤≤  ≤ |xk |+|yk |.
xk + yk2  x2k + yk2 x2k + yk2  x2 + y 2 
k k x2k + yk2 x2k + yk2

We have that
lim (xk , yk ) = (0, 0) =⇒ lim |xk | + |yk | = 0.
k→∞ k→∞

Then, according to the sandwich theorem

x3 + y 3
0 ≤ lim ≤ lim |xk | + |yk | = 0.
(x,y)→(0,0) x2 + y 2 k→∞

Hence,
x3 + y 3
lim = 0.
(x,y)→(0,0) x2 + y 2

Excercise 5: Study
x3 + y 3
lim .
(x,y)→(0,0) xy

Solution:

Obviously, the function for which we study the limit is

2 x3 + y 3
f : R \{(0, 0)} f (x, y) = .
xy
Once again we make use of the sequential characterization of limits. We consider, to begin
with, a random sequence (xk ) ⊆ R \{0} such that limk→∞ xk = 0.
Just like in the case of Exercise 3, we make use of this sequence, in order to generate
(yk ) ⊆ R \{0}, with the help of a m ∈ R∗ . Thus

yk = m · xk , ∀k ∈ N.

Then
x3k + yk3 x3 + m3 x3k 1 + m3
= k = xk .
xk y k x k · m · xk m
This means that
1 + m3 1 + m3
lim f (xk , yk ) = lim xk = · 0 = 0. (1)
k→∞ k→0 m m

4
 But this does not mean that the limit of the function exists in general, because... in
order to be so, it has to be 0 for all the sequences that tend to (0, 0).
Consider now another sequence, with the general term

zk = m · x2k , ∀k ∈ N.

Then
lim (xk , zk ) = (0, 0)
k→∞

and
x3k + m3 · x6k 1 + m3 x3k 1
lim f (xk , zk ) = lim 2
= lim = . (2)
k→∞ k→∞ xk · mxk k→∞ m m
From (1) and (2) we have
1
lim f (xk , yk ) = 0 6= = lim f (xk , zk ).
k→∞ m k→∞
Thus, the limit does not exist.

Excercise 5: Consider the set

A = {(x, y) ∈ R2 : xy 6= 0}

and the function

1 1
f : A → R, f (x, y) = x sin + y sin , ∀(x, y) ∈ A.
y x
Notice that (0, 0) ∈ A0 . Prove that
a) ∀y ∈ R, 6 ∃ limx→0 f (x, y).
b) ∃ lim f (x, y).
(x,y)→(0,0)

Solution:
a) Choose y ∈ R∗ randomly. We will prove that 6 ∃ limx→0 f (x, y), with the help of
sequences. Since we have the function sin in the expression of the function, we make use
of its particular values.
Consider the sequences (xk ) ⊆ R, and (zk ) ⊆ R with the general terms
1 1
xk = , , zk := π ∀k ∈ N.
2kπ 2kπ + 2

Then
lim xk = lim zk = 0,
k→∞ k→∞

5
but
1 1 1
lim f (xk , y) = lim sin + y sin(2kπ) = 0 · sin + y · 0 = 0,
k→∞ k→∞ 2kπ y y
while
1 1  π 1
lim f (zk , y) = lim π sin + y sin 2kπ + = 0 · sin + y · 1 = y.
k→∞ k→∞ 2kπ + y 2 y
2

Since y 6= 0 it follows that

lim f (xk , y) 6= lim f (zk , y).

k→∞ k→∞

Thus
6 ∃ lim f (x, y).
x→0

Recall the fact that y was randomly chosen, therefore, statement a) is proved.

b) We prove that
∃ lim f (x, y),
(x,y)→(0,0)

by using the sandwich theorem.

Let x, y ∈ R∗ be randomly chosen then, the following chain of inequalities is satisfied:
     
 1 1   1  1 
0 ≤ x sin + y sin  ≤ |x| · sin  + |y| · sin  ≤ |x| + |y|.
y x y x
Notice that lim(x,y)→(0,0) |x| + |y| = 0 + 0 = 0, hence, due to the sandwich theorem,

lim f (x, y) = 0.
(x,y)→(0,0)

Excercise 6: Consider the set

A = {(x, y) ∈ R2 : y > 0}

and the function

f :A→R by f (x, y) = x ln y.

Note that (0, 0) ∈ A0 . Study

lim f (x, y).

(x,y)→(0,0)

Solution:
Consider a random sequence (yk ) ⊂ (0, ∞) such that limk→∞ yk = 0.

6
 Define for each k ∈ N,
1
xk := .
ln yk
Then
1
lim xk = = 0.
k→∞ −∞
We are interesting in determining

lim f (xk , yk )
k→∞

Before considering the limit, we compute for a random k ∈ N

1
f (xk , yk ) = · ln yk = 1.
ln yk
So, (f (xk , yk ))k∈N is actually the constant sequence 1, therefore

lim f (xk , yk ) = 1
k→∞

Define for each k ∈ N,

1
tk := 2 .
ln yk
Then
1
lim xk = = 0.
k→∞ ∞
We are interesting in determining

lim f (tk , yk )
k→∞

Before considering the limit, we compute for a random k ∈ N

1 1 1
f (tk , yk ) = 2 · ln yk = lim = =0
ln yk k→∞ ln yk −∞

So, (f (tk , yk ))k∈N is actually the constant sequence 1, therefore

lim f (tk , yk ) = 0
k→∞

.
Thus

lim (xk , yk ) = lim (tk , yk ) = 0, but lim f (xk , yk ) = 1 6= 0 = lim f (tk , yk ).

k→∞ k→∞ k→∞ k→∞

Thus
lim f (x, y) does not exist.
(x,y)→(0,0)

For further details, take a look at the solved exercises on

Seminar4-ttrif