CHAPTER 7

Continuous Functions in Euclidean Spaces

1. Continuous Functions
Definition 1.1. Let (X; ρ) and (Y ; η) be two metric spaces, f : X → Y a function
and x0 ∈ X.
• f is continuous at x0 if for all  > 0 there exists δ > 0 such that, whenever x ∈ X
and ρ(x, x0 ) < δ then η(f (x), f (x0 )) < .
• f is continuous if f is continuous at all x0 ∈ X.
Remark 1.2. We may consider the ”more general” setting
f : D (⊆ X) → Y, x0 ∈ D,
with the following definition: f is continuous at x0 if for all  > 0 there exists δ > 0
such that, whenever x ∈ D and ρ(x, x0 ) < δ, then η(f (x), f (x0 )) < .
But this is not more general than the definition provided before since it coincides
with the case when D is considered as a metric space with the metric ρ|D×D .
Fact 1.3. Let (X; ρ) and (Y ; η) be two metric spaces, f : X → Y a function, and
x0 ∈ X. TFAE:
(i) f is continuous at x0 .
(ii) For any  > 0 there exists δ > 0 such that f (Bδ (x0 )) ⊆ B (f (x0 )).
(iii) For any U ∈ V(f (x0 )) there exists V ∈ V(x0 ) such that f (V ) ⊆ U .
(iv) For any open set U ⊆ Y such that f (x0 ) ∈ U there exists an open set V ⊆ X
such that x0 ∈ V and f (V ) ⊆ U .
Proof. Exercise. 
The item (iv) from the previous statement shows that continuity is a purely topo-
logical concept and hence it enables us to obtain a general concept of continuity for
functions between topological spaces.
Definition 1.4. Let (X; T ) and (Y ; S) be topological spaces. Let f : X → Y be
a function and x0 ∈ X. Then, f is continuous at x0 if, for any U ∈ S such that
f (x0 ) ∈ U there exists V ∈ T such that x0 ∈ V and f (V ) ⊆ U .

Continuity is related to the more general concept of limit.

Definition 1.5. Let (X; ρ) and (Y ; η) be metric spaces, D ⊆ X. Let f : D → Y
be a function and x0 ∈ X and y0 ∈ Y . Then, f has limit y0 at x0 if:
167
168 7. CONTINUOUS FUNCTIONS IN EUCLIDEAN SPACES

• x0 is an accumulation point for D.

• For any  > 0 there exists δ > 0 such that, whenever x ∈ D \ {x0 } and
ρ(x, x0 ) < δ, then η(f (x), y0 ) < .
Remark 1.6. If f has limit y0 at x0 then y0 is unique (prove it), hence we can
denote:
lim f (x) = y0 .
x→x0

In the following we consider two metric spaces (X; ρ) and (Y ; η), D ⊆ X a nonempty
set, and a function f : D → Y .
Fact 1.7. Let x0 ∈ D. Then:
(i) If x0 is isolated in D then f is continuous at x0 .
(ii) If x0 is an accumulation point for D then f is continuous at x0 if and only if
limx→x0 f (x) = f (x0 ).

Proof. Exercise. 
Fact 1.8 (Sequential Characterization of Limits of Functions). Assume that x0 ∈ X
is an accumulation point for D and let and y0 ∈ Y . Then, the following assertions are
equivalent:
(i) limx→x0 f (x) = y0 .
(ii) For any sequence (xn )n≥1 , with xn ∈ D and xn 6= x0 for all n ∈ N and
limn→∞ xn = x0 , it follows that limn→∞ f (xn ) = y0 .

Proof. ”(i)⇒(ii)”. For any  > 0 there exists δ > 0 such that, whenever x ∈
D \ {x0 } and ρ(x, x0 ) < δ then η(f (x), y0 ) < . Take (xn )n≥1 a sequence in D such
that xn 6= x0 for all n and limn→∞ xn = x0 . Then, there exists Nδ ∈ N such that,
for all n ∈ N, if n ≥ Nδ then ρ(xn , x0 ) < δ, hence η(f (xn ), y0 ) < . This means that
limn→∞ f (xn ) = y0 .
”(ii)⇐(i)” Assume that, for any sequence (xn )n≥1 with all elements in D \ {x0 } and
limn→∞ xn = x0 we have limn→∞ f (xn ) = y0 . By contradiction, let us assume that f (x)
does not converge to y0 as x approaches x0 . Then, there exists 0 > 0 such that, for
any δ > 0 there exists x ∈ D \ {x0 } with ρ(x, x0 ) < δ and η(f (x), y0 ) ≥ 0 . For any
n ∈ N, take δ = n1 > 0 hence, there exists xn ∈ D \ {x0 } with ρ(xn , x0 ) < δ = n1 and
ρ
η(f (xn ), y0 ) ≥ 0 . This means that xn −−−→ x0 but f (xn ) does not converge to y0 with
n→∞
respect to η. 

In a similar fashion we get the following statement.

Fact 1.9 (Sequential Characterization of Continuity). Assume that x0 ∈ D. Then,
the following assertions are equivalent:
(i) f is continuous at x0 .
ρ η
(ii) For any sequence (xn )n≥1 in D such that xn →
− x0 we have f (xn ) →
− y0 .
n n
 1. CONTINUOUS FUNCTIONS 169

Fact 1.10 (Continuity of Composition of Continuous Functions). Let (X; ρ), (Y ; η)

and (Z; θ) be three metric spaces and consider two functions f : X → Y and g : Y → Z.
Assume that, given x0 ∈ X, f is continuous at x0 and that g is continuous at f (x0 ) ∈ Y .
Then g ◦ f is continuous at x0 .
Proof. Consequence of the sequential characterization of continuity. 
Fact 1.11 (Limits of Composition of Functions). Let (X; ρ), (Y ; η) and (Z; θ) be
three metric spaces and consider two functions f : F → Y and g : G → Z, subject to
the following assumptions:
• x0 ∈ X is an accumulation point for F ⊆ X.
• limx→x0 f (x) = y0 ∈ Y .
• y0 ∈ G ⊆ Y and g is continuous at y0 .
• x0 is an accumulation point for the domain of g ◦ f , that is, the set {x ∈ F |
f (x) ∈ G} = f −1 (G).
Then,
(1.1) lim (g ◦ f )(x) = g(y0 ).
x→x0

Proof. Firstly, since x0 is an accumulation point for the domain of g◦f , the limit in
(1.1) makes sense. Then, in order to prove (1.1), we use the sequential characterization
of limits, so let (xn )n be an arbitrary sequence in f −1 (G) \ {x0 } that converges to
η
x0 , hence the sequence f (xn ) →− y0 ∈ G. Since g is continuous at y0 it follows that
n
θ θ
g(f (xn )) →
− g(y0 ), that is, (g ◦ f )(xn )) →
− g(y0 ). By the sequential characterization of
n n
limits, we conclude that (1.1) holds true. 
Fact 1.12 (Functions Between Euclidean Spaces). Let D ⊆ Rp and f : D → Rq
be a function, hence f (x) = (f1 (x), . . . , fq (x)) for all x ∈ D, where fj : D → R is a
function for all j = 1, . . . , q.
(i) Let x0 ∈ D0 , i.e. x0 is an accumulation point for D, and let y (0) ∈ Rq , hence
(0) (0)
y (0) = y1 , . . . , yq ∈ Rq . Then,
(0)
lim f (x) = y (0) if and only if, for all j = 1, . . . , q, lim fj (x) = yj .
x→x0 x→x0

(ii) Let x0 ∈ D. Then,

f is continuous at x0 if and only if, for all j = 1, . . . , q, fj is continuous at x0 .
Proof. (i) We assume that limx→x0 f (x) = y (0) and use the norm k · k∞ . Thus,
for any  > 0 there exists δ > 0 such that, for all x ∈ D \ {x0 }, if kx − x0 k < δ then
kf (x) − y (0) k∞ < .
For arbitrary j ∈ {1, . . . , q} we have
(0)
|fj (x) − yj | ≤ kf (x) − y (0) k∞ < ,
which means that fj is continuous at x0 .
170 7. CONTINUOUS FUNCTIONS IN EUCLIDEAN SPACES

Conversely, assume that, for any j = 1, . . . , q, we have

(0)
lim fj (x) = yj .
x→x0

Then, for any  > 0 there exists δj > 0 such that, for any x ∈ D\{x0 }, if kx−x0 k∞ < δj
(0)
then |fj (x) − yj | < . Take δ := min{δ1 , . . . , δq } > 0. Then, for any j = 1, . . . , q, if
(0)
kx − x0 k∞ < δ ≤ δj then |fj (x) − yj | <  hence
(0)
kf (x) − y (0) k∞ = max |fj (x) − yj | < .
j=1,...,q

(ii) We distinguish two case: if x0 isolated, nothing to prove. If x0 is an accumulation

point for D, we use (i). 
Example 1.13. (1) Let the function f : R2 → R be defined by
(
xy
2 2, (x, y) 6= (0, 0)
f (x, y) = x +y
0, (x, y) = (0, 0).
Then, f is continuous on R2 \ {(0, 0)}, (0, 0) is an accumulation point for R2 , but f
does not have a limit when (x, y) → (0, 0). Indeed, to see the latter fact, we observe
that:
when x = 0 and y → 0 we have f (0, y) = 0 → 0,
when y = 0 and x → 0 we have f (x, 0) = 0 → 0,
1 1
when x = y → 0 we have f (x, y) = → .
2 2
(2) Let the function f : R2 → R be defined by
( xy
√ , (x, y) 6= (0, 0)
f (x, y) = x2 +y 2
0, (x, y) = (0, 0).
Then, f continuous on R2 . Indeed, at (x0 , y0 ) 6= (0, 0) is clear, while, for the continuity
at (0, 0), we observe that
p
|xy| x2 + y 2 k(x, y)k2
|f (x, y)| = p ≤ = .
x2 + y 2 2 2
Definition 1.14. (1) A curve in R2 is a continuous function γ : I → Rq , q ≥ 1
where I is an interval.
(2) If the interval I = [a, b] is compact, then the curve has endpoints x = γ(a) and
y = γ(b). In this case we say that γ is a path joining x and y.
(3) A curve with endpoints x and y is called closed if x = y.
Example 1.15. (1) γ(t) = (cos t, sin t), γ : [0, 2π] → R2 , γ(0) = γ(2π) so γ is a
closed curve.
(2) γ(t) = (t2 , t3 ), γ : [0, 1] → R2 , γ(0) = (0, 0) and γ(1) = (1, 1), is a curve with
endpoints but not closed.
 2. PROPERTIES OF CONTINUOUS FUNCTIONS 171

(3) γ(t) = (t cos t, t sin t, t), γ : R → R3 , is a curve with no endpoints. It is a spiral

on the cone {(x, y, z) | x2 + y 2 = |z|}.
Definition 1.16. A surface in Rd (q ≥ 2) is a continuous function F : A → Rd ,
such that, for some D open and nonempty set in R2 , we have D ⊆ A ⊆ D.
Example 1.17. (1) 2-Dimensional Sphere in R3 .
h π πi
F : [0, 2π) × − , → R3 ,
2 2
h π πi
θ ∈ [0, 2π), φ ∈ − ,
F (θ, φ) = (cos θ cos φ, sin θ cos φ, sin φ), .
2 2
(2) 2-Dimensional Semi Sphere in R3 . G : B → R3 defined by
 p 
G(x, y) = x, y, 1 − x2 − y 2 , (x, y) ∈ B,
where,
B = (x, y)|x2 + y 2 ≤ 1 .


2. Properties of Continuous Functions

Theorem 2.1 (Topological Characterization of Continuity). Let (X; ρ), (Y ; η) be
metric spaces and f : X → Y a function. TFAE:
(i) f is continuous.
(ii) For all U open in Y , f −1 (U ) is open in X.
(iii) For all F closed in Y , f −1 (F ) is closed in X.
Proof. (i)⇒(ii)). Let U open in Y and x ∈ f −1 (U ), i.e. f (x) ∈ U . Then there
exists  > 0 such that B (f (x)) ⊆ U . Since f is continuous at x, there exists δ > 0 such
that f (Bδ (x)) ⊆ B (f (x)) ⊆ U , hence Bδ (x) ⊆ f −1 (B (f (x))) ⊆ f −1 (U ), i.e. f −1 (U ) is
open.
((ii)⇒(i)). Let x ∈ X and  > 0. Then B (f (x)) is open in Y , hence f −1 (B (f (x)))
is open in X. Since x ∈ f −1 (B (f (x))) it follows that there exists δ > 0 such that
Bδ (x) ⊆ f −1 (B (f (x))), i.e. f (Bδ (x)) ⊆ B (f (x)). We have proven that f is continuous
at each x ∈ X.
((ii)⇔(iii). Since f −1 (Y \ A) = f −1 (Y ) \ f −1 (A) = X \ f −1 (A). 
Corollary 2.2. Let (X; ρ) and (Y ; η) be two metric spaces and let ∅ 6= D ⊆ X
and f : D → Y be a function. TFAE:
(i) f is continuous.
(ii) For all U open in Y , f −1 (U ) is relatively open in D.
(iii) For all F closed in Y , f −1 (F ) is relatively closed in D.
6 D ⊆ Rp be open, f : D → Rq a function. TFAE:
Corollary 2.3. Let ∅ =
(i) f is continuous.
(ii) For all U open in Rq , f −1 (U ) is open in Rp .
(iii) For all F closed in Rq , f −1 (F ) is closed in Rp .

any cont func pushbacks open sets to open sets

172 7. CONTINUOUS FUNCTIONS IN EUCLIDEAN SPACES

Theorem 2.4. Let (X; ρ), (Y ; η) be metric spaces, f : X → Y a continuous func-

tion and K a compact set in X. Then f (K) is compact.

Proof. Let {Ui | i ∈ J } be an open (in Y ) covering of f (K), that is: for all i ∈ J ,
the set Ui is open in Y and
[
f (K) ⊆ Ui .
i∈J
−1
Then, for all i ∈ J , f (Ui ) is open in X and
[ [
K ⊆ f −1 (f (K)) ⊆ f −1 ( Ui ) = f −1 (Ui ),
i∈J i∈J
pullback commutes with all set operations
hence {f −1 (Ui )|i ∈ J } is an open covering of K.
Since K is compact, there exists i1 , . . . , in ∈ J such that
n
[ n
[
−1 −1
K⊆ f (Uik ) = f ( Uik ),
k=1 k=1 (heine borell definition of compactness
union of pullbacks for any open cover we get a finite subcover)
hence
n
[ n
[
−1
f (K) ⊆ f (f ( Uik )) ⊆ Uik ,
pushforward of the pulback k=1 k=1
hence {Uik | k = 1, . . . , n} is a finite subcover. smaller then original set 
Corollary 2.5. Let f : X → R be a continuous function and K compact and
nonempty set in X. Then, f is bounded on K, i.e. f (K) is bounded in R, and the
extreme values of f on K are attained, i.e. there exists xm , xM ∈ K such that
f (xm ) = inf f, f (xM ) = sup f.
K K

Proof. f (K) is compact in R, hence closed and bounded. Since f (K) bounded
⇒, we have inf K f, supK f ∈ R. But, inf K f ∈ f (K) = f (K), hence, there exists
xm ∈ K such that inf K f = f (xm ). Similarly, supK f ∈ f (K) = f (K), hence there
exists xM ∈ K such that supK f = f (xM ). 
Corollary 2.6. Let f : X → Rq be a continuous function. Then, for any K
nonempty and compact subset of X, there exists xm , xM ∈ K such that
kf (xm )k = inf kf k, kf (xM )k = sup kf k.
K K

Proof. Indeed, we have a continuous function, k · k : Rq → R, and then a compo-

sition of two continuous functions,
kf k = k · k ◦ f : X → R. 
Theorem 2.7. Let (X; ρ), (Y ; η) be metric spaces, f : X → Y a continuous func-
tion, and C a connected set in X. Then f (C) is connected in Y .
 2. PROPERTIES OF CONTINUOUS FUNCTIONS 173

Proof. By contrapositive, assume that f (C) is separated hence, there exist U , V

open in Y such that f (C) ⊆ U ∪ V , f (C) ∩ U 6= ∅, f (C) ∩ V 6= ∅, and ∩U ∩ V = ∅.
Then f −1 (U ), f −1 (V ) are open in X. In addition,
C ⊆ f −1 (f (C)) ⊆ f −1 (U ∪ V ) = f −1 (U ) ∪ f −1 (V ),
and
f −1 (U ) ∩ f −1 (V ) = f −1 (U ∩ V ) = f −1 (∅) = ∅.
We prove that C ∩ f −1 (U ) 6= ∅. Indeed, since f (C) ∩ U 6= ∅, there exists y ∈ f (C)
and y ∈ U hence there exists x ∈ C such that f (x) = y ∈ U , hence x ∈ f −1 (U ), i.e.
x ∈ C ∩ f −1 (U ).
Similarly we prove that C ∩ f −1 (V ) 6= ∅.
Thus, f −1 (U ) and f −1 (V ) separate C. 
Corollary 2.8. (1) Let f : (X; ρ) → R be a continuous function, C connected in
X. Then f (C) is an interval.
(2) Let I be an interval in R and γ : I → Rd continuous (a curve). Then γ(I) is
connected.
Definition 2.9. A subset S ⊆ (X; ρ) is called pathwise connected if for all a, b ∈ S
there exists a (continuous) curve γ : [0, 1] → S such that a = γ(0) and b = γ(1).
Fact 2.10. If S is pathwise connected, then it is connected.

Proof. Assume S is not connected, let U , V open in X and separating S. Then

there exist a ∈ S ∩ U and b ∈ S ∩ V . Since S is pathwise connected, there exists
γ : [0, 1] → S continuous such that γ(0) = a and γ(1) = b. But then U and V separate
γ([0, 1]), contradiction! 
Example 2.11. A set in R2 that is connected but not pathwise connected.
1 1 1
S = {(0, y)| − 1 ≤ y ≤ 1} ∪ {(x, sin )| − < x < , x 6= 0}.
x π π
Fact 2.12. Assume that D is open in Rd . Then D is connected iff D is pathwise
connected.

Proof. ”⇐”. Holds in general.

p
”⇒”. On D we define a relation: x ∼ y if there exists γ : [0, 1] → D continuous
such that γ(0) = x and γ(1) = y.
p
• ∼ is an equivalence relation on D.
• For all x ∈ D its equivalence class [x]p is an open set.
p
If D is not pathwise connected then there exist at least two different cosets w.r.t. ∼,
hence D is disconnected. 
174 7. CONTINUOUS FUNCTIONS IN EUCLIDEAN SPACES

3. Uniform Continuity

Definition 3.1. Let (X; ρ) and (Y ; η) be two metric spaces. A function f : X → Y

is uniformly continuous if, for any  > 0 there exists δ > 0 such that, for arbitrary
x1 , x2 ∈ X, if ρ(x1 , x2 ) < δ then η(f (x1 ), f (x2 )) < .
Theorem 3.2. Let (X; ρ and (Y ; η) be metric spaces. If the function f : X → Y is
continuous and X is compact, then f is uniformly continuous.

Proof. Let  > 0. Since f is continuous on X, for all x ∈ X there exists δx > 0
such that, for all z ∈ X with ρ(x, z) < δx we have η(f (x), f (z)) < /2.
Since {Bδx /2 (x)|x ∈ X} is an open covering of X, which is compact, it follows that
there exist x1 , . . . , xn ∈ X such that
n
[
X⊆ Bδxi /2 (xi ).
i=1

δxi
Let δ := min{ 2
|i = 1, . . . , n} > 0 and let x, z ∈ X such that ρ(x, z) < δ. Then
δxj
there exists j ∈ {1, . . . , n} such that x ∈ Bδxj /2 (xj ) i.e. ρ(x, xj ) < 2
. Then,
δx j δx δx
ρ(z, xj ) ≤ ρ(z, x) + ρ(x, xj ) < δ + ≤ j + j = δxj ,
2 2 2

hence η(f (x), f (xj )) < 2
and η(f (z), f (xj )) < 2 . Then
η(f (x), f (z)) ≤ η(f (x), f (xj )) + η(f (z), f (xj )) < . 
Theorem 3.3. Let (X; ρ) be a compact metric space, (Y ; η) be a complete metric
space, ∅ =
6 D ⊆ X and f : D → Y a function. TFAE:
(i) f is uniformly continuous on D.
(ii) There exists f : D → Y such that f |D = f and f is continuous on D.

Proof. (ii)⇒(i). D closed in X and X is compact, hence D is compact and f is

uniformly continuous on D, hence f = f |D is uniformly continuous on D.
(i) ⇒(ii). Assume that f is uniformly continuous.
We first show that f maps Cauchy sequences from D to Cauchy sequences in Y .
Indeed, let  > 0. Then there exists δ > 0 such that, for all x, z ∈ D, if ρ(x, z) < δ
then η(f (x), f (z)) < . Let (xn )n≥1 be a ρ-Cauchy sequence in D. Then, for all  > 0,
with δ > 0 as before, there exists N ∈ N such that, for all n ∈ N, if m, n ≥ N then
ρ(xm , xn ) < δ, hence η(f (xm ), f (xn )) < .
ρ
Let x ∈ D be arbitrary. Then there exists (xn )n≥1 in D such that xn →
− x, hence
n
(xn )n≥1 is ρ-Cauchy, hence (f (xn ))n≥1 is η-Cauchy. Since (Y ; η) is complete, there
η
exists yx ∈ Y such that f (xn ) →
− yx . Define f (x) := yx .
n
 4. SEQUENCES AND SERIES OF FUNCTIONS 175

We show that the definition of f is correct. To this end, we use the interlacing
ρ
method: If (xn )n≥1 is a sequence in D such that xn → − x and (zn )n≥1 is another
n
ρ
sequence in D such that zn →
− x, then the sequence (tn )n≥1 defined by
n

x n2 , if n = 2k
tn =
z n−1 , if n = 2k + 1
2

ρ η η η
is a sequence in D and tn →
− x. Then f (xn ) →
− yx , f (zn ) →
− zx , f (tn ) →
− tx . Since
n n n n
(f (xn ))n≥1 and (f (zn ))n≥1 are subsequences of (f (tn ))n≥1 , it follows that yx = tx = zx .
We show that f is continuous on D. To this end, we use the sequential characteri-
ρ
zation of continuity. Let x ∈ D and (xn )n≥1 a sequence in D such that xn →
− x. Then,
n
after a moment of thought we see that, for all n ≥ 1 there exists xn ∈ D such that
ρ(xn , xn ) < n1 and η(f (xn ), f (xn )) < n1 . Consequently,
1
ρ(xn , x) ≤ ρ(xn , xn ) + ρ(xn , x) < + ρ(xn , x) →
− 0,
n n
ρ η
hence xn →
− x and then, by the definition of f , we have f (xn ) →
− f (x), and then
n n

1
η(f (xn ), f (x)) ≤ η(f (xn ), f (xn )) + η(f (xn ), f (x)) < + η(f (xn ), f (x)) →
− 0
n n
η
hence f (xn ) →
− f (x). 
n

Corollary 3.4. Let (X; ρ) be a metric space, (Y ; η) be a complete metric space,

∅=6 D ⊆ X and f : D → Y a function. If f is uniformly continuous on D then there
exists and it is unique a function f : D → Y such that f |D = f and f is continuous on
D.

Proof. The statement follows from the observation that in the proof of the impli-
cation (i)⇒(ii) of the previous theorem, we do not use the compactness of the metric
space (X; ρ) at all. 

4. Sequences and Series of Functions

Let X be a nonempty set and (Y ; η) a metric space. We consider sequences (fn )n≥1
of functions such that fn : X → Y for all n ∈ N.
Definition 4.1. (fn )n≥1 converges pointwise to f : X → Y if, for all x ∈ X and
for all  > 0 there exists N ∈ N such that, for all n ≥ N we have η(f (x), fn (x)) < ,
i.e. for all x ∈ X, (fn (x))n≥1 converges to f (x) in Y with respect to η.
(fn )n≥1 converges uniformly to f : X → Y if for all  > 0 there exists N ∈ N such
that for all n ≥ N and for all x ∈ X we have η(f (x), fn (x)) < .
176 7. CONTINUOUS FUNCTIONS IN EUCLIDEAN SPACES

Fact 4.2. (1) If there exists (αn )n≥1 such that αn ≥ 0 for all n ∈ N and αn →
− 0
n
unif.
and η(f (x), fn (x)) ≤ αn for all n ∈ N and ∀x ∈ X, then fn −−→ f .
n
(2) If (X; ρ) and (Y ; η) are metric spaces and we have a sequence of continuous
unif.
functions fn : X → Y , n ∈ N and a function f : X → Y such that fn −−→ f , then f is
n
continuous on X.
Proof. Let x0 ∈ X and  > 0. There exists N ∈ N such that for all n ≥ N and
all x ∈ X

η(fn (x), f (x)) < .
3
fN is continuous at x0 , so there exists δ > 0 such that for all x ∈ X, ρ(x, x0 ) < δ
implies η(fN (x), fN (x0 )) < 3 . Then for all x ∈ X, if ρ(x, x0 ) < δ then
η(f (x), f (x0 )) ≤ η(f (x), fN (x)) + η(fN (x), fN (x0 )) + η(fN (x0 ), f (x0 ))
  
< + + = . 
3 3 3
Example 4.3. (1) Let B1 (0) be the open unit ball in Rd and let the sequence of
functions fn : B1 (0) → R be defined by fn (x) = kxkn2 . Let

0, x ∈ B1 (0),
f (x) =
1, x ∈ ∂B1 (0).
pointwise
Then fn −−−−−→ f but not uniformly.
n
(2) Consider B1 (0) ∈ R2 , the sequence of functions fn : B1 (0) → R2 defined by
 2
x1 − nx22

nx1
fn (x1 , x2 ) = ,
1 + nx22 1 + nx21
and the function f : B1 (0) → R2 defined by
(−1, x11 ),


 x1 6 0, x2
= 6 0,
=
(−1, 0), x1 = 0, x2 6= 0,

f (x) = 2 1
(x , ), x1 6= 0, x2 = 0,
 1 x1


(0, 0), x1 = 0, x2 = 0.
pointwise
Then fn −−−−−→ f but not uniformly.
n

Definition 4.4. Let fn : X → Y , n ∈ N, for X 6= ∅ and (Y ; η) a metric space.

(fn )n≥1 is uniformly Cauchy if for all  > 0 there exists N ∈ N such that for all
m, n ≥ N we have
η(fn (x), fm (x)) < .
unif.
Fact 4.5. (1) If fn −−→ f then (fn )n≥1 is uniformly Cauchy.
n
(2) If (Y ; η) is complete and the sequence (fn )n≥1 is uniformly Cauchy then there
unif.
exists f : X → Y such that fn −−→ f .
n
 4. SEQUENCES AND SERIES OF FUNCTIONS 177

Proof. (1) Exercise.

(2) Let  > 0. There exists N ∈ N such that for any m, n ≥ N and any x ∈ X we
have

(∗) η(fm (x), fn (x)) < .
2
Then, for each x ∈ X the sequence (fn (x))n≥1 is Cauchy in Y , which is complete, hence
there exists f (x) ∈ Y such that fn (x) →
− f (x). Due to the uniqueness of the limit in
n
pointwise
Y , there exists a function f : X → Y such that fn −−−−−→ f .
n
Letting m → +∞ in (∗), for all n ≥ N and all x ∈ X, we have

η(f (x), fn (x)) ≤ < ,
2
unif.
hence fn −−→ f . 
n

Remark 4.6. For all z ∈ Y the map Y 3 y 7→ η(y, z) ∈ R is continuous.

Definition 4.7. Let X 6= ∅ and let (V, k · k) be a normed space.
B(X; V ) := {f : X → V | f is a bounded function} .
For all f ∈ B(X; V ), let kf kX := supkf (x)k.
x∈X

Fact 4.8. We use the notation as in the previous definition.

(1) B(X; V ) is a vector space and k · kX is a norm, called the supremum norm on
X.
(2) If (V ; k · k) is complete, then the normed space (B(X; V ); k · kX ) is complete.

Proof. (1) Exercise.

(2) Let (fn )n≥1 be a Cauchy sequence in B(X; V ) with respect to the norm k · kX ,
that is, for any  > 0 there exists N ∈ N such that, for all m, n ≥ N we have
kfm − fn kX < 2 , hence, for all x ∈ X

(∗) kfm (x) − fn (x)k ≤ kfm − fn kX < .
2
Then, for all x ∈ X the sequence (fn (x))n≥1 is Cauchy in V , which is complete, hence
there exists f (x) ∈ V such that
kfn (x) − f (x)k −−−→ 0.
n→∞

After a moment of thought we see that X 3 x 7→ f (x) ∈ V is a function.

We show that f ∈ B(X; V ) and kfn − f kX →− 0. Indeed, for all x ∈ X let m → ∞
n
in (∗) hence

(∗∗) kf (x) − fn (x)k ≤ < ,
2
178 7. CONTINUOUS FUNCTIONS IN EUCLIDEAN SPACES

and then

(∗ ∗ ∗) sup kf (x) − fn (x)k ≤ < .
x∈X 2

Also, for all x ∈ X, letting n = N in (∗∗), we have


kf (x)k ≤ kf (x) − fN (x)k + kfN (x)k ≤ + kfN kX < +∞
2
hence f ∈ B(X; V ).
k·kX
On the other hand, from (∗ ∗ ∗), for all n ≥ N kf − fn kX < , hence fn −−→ f . 
n

Corollary 4.9. Let (X; ρ) be a compact metric space, (V ; k · k) a Banach space

and
C(X; V ) := {f : X → V |f continuous} .
Then, C(X; V ) is a vector subspace of B(X; V ) as a vector subspace and (C(X; V ); k·kX )
is closed, hence a complete normed space, i.e. a Banach space.

Proof. This follows from the previous fact, the observation that convergence
(Cauchy) with respect to the norm k · kX is the same with uniform convergence
(Cauchy), and the fact that, if (fn )n is a sequence of continuous functions that con-
verges to a function f then the function f is continuous. 
Definition 4.10. Let ∅ 6= X be a set, let (V, k · kV ) be a normed space and let
(fn )n≥1 be a sequence of functions fn : X → V .
A formal sum ∞
P
n=1 fn is called a series of functions on X and valued in V . Letting,
for all n ∈ N,
X n
sn = fn ,
k=1

(sn )∞ is the sequence of partial sums of the series ∞

P
n=1 n=1 fn .
P∞
The series n=1 fn pointwise converges if the sequence (sn )∞ n=1 converges pointwise
to a function f : X → V .
The series ∞ ∞
P
n=1 fn uniformly converges if the sequence (sn )n=1 converges uniformly
to a function f : X → V .
P∞ P∞
The series n=1 f n absolutely converges if n=1 kfn kV converges. This can be
pointwise or uniformly.
Fact 4.11. With notation as before, assume that the normed space (V ; k · k) is
complete.
(1) If ∞
P
n=1 fn absolutely converges then it converges. P∞
(2) (Weierstraß
P∞ Test) If kf n (x)k ≤ αn for all x ∈ X, n ∈ N, and n=1 αn < +∞
then n=1 fn converges absolutely and uniformly.
 5. EXERCISES 179

5. Exercises

1. Let (X; ρ) and (Y ; η) be two metric spaces, f : X → Y a function, and x0 ∈ X.

Show that the following assertions are equivalent:
(i) f is continuous at x0 .
(ii) For any  > 0 there exists δ > 0 such that f (Bδ (x0 )) ⊆ B (f (x0 )).
(iii) For any U ∈ V(f (x0 )) there exists V ∈ V(x0 ) such that f (V ) ⊆ U .
(iv) For any open set U ⊆ Y such that f (x0 ) ∈ U there exists an open set V ⊆ X
such that x0 ∈ V and f (V ) ⊆ U .
2. We consider two metric spaces (X; ρ) and (Y ; η), D ⊆ X a nonempty set, and
a function f : D → Y . Let x0 ∈ D. Show that:
(i) If x0 is isolated in D then f is continuous at x0 .
(ii) If x0 is an accumulation point for D then f is continuous at x0 if and only if
limx→x0 f (x) = f (x0 ).
3. With notation as in the previous exercise, assume that x0 ∈ X is an accumulation
point for D and let and y0 ∈ Y . Then, the following assertions are equivalent:
(i) limx→x0 f (x) = y0 .
(ii) For any sequence (xn )n≥1 , with xn ∈ D and xn 6= x0 for all n ∈ N and
limn→∞ xn = x0 , it follows that limn→∞ f (xn ) = y0 .
4. Let (X; ρ and (Y ; η) be two metric spaces and let ∅ 6= D ⊆ X and f : D → Y
be a function. Show that the following assertions are equivalent:
(i) f is continuous.
(ii) For all U open in Y , f −1 (U ) is relatively open in D.
(iii) For all F closed in Y , f −1 (F ) is relatively closed in D.
6 D ⊆ Rp be open and let f : D → Rq be a function. TFAE:
5. Let ∅ =
(i) f is continuous.
(ii) For all U open in Rq , f −1 (U ) is open in Rp .
(iii) For all F closed in Rq , f −1 (F ) is closed in Rp .
6. Assume that X is compact and let f, g : X → R be two continuous functions
such that g(x) 6= 0 for all x ∈ X. Show that f /g is continuous.
7. Consider the set in R2 defined by
1 1 1
S = {(0, y)| − 1 ≤ y ≤ 1} ∪ {(x, sin )| − < x < , x 6= 0}.
x π π
Show that S is connected but not pathwise connected.
8. Let A be a compact subset of Rd and define B as follows:
B = {ta | a ∈ A, t ∈ [0, 1]}.
Investigate whether B is a compact set or not.
180 7. CONTINUOUS FUNCTIONS IN EUCLIDEAN SPACES

9. Show that the set in R2 defined by

1 1 1
S = {(0, y)| − 1 ≤ y ≤ 1} ∪ {(x, sin )| − < x < , x 6= 0}
x π π
that is connected but not pathwise connected.
10. Let fn : X → Y , n ∈ N, for X 6= ∅ and (Y ; η) a metric space. Show that:
unif.
(1) If fn −−→ f then (fn )n≥1 is uniformly Cauchy.
n
(2) If (Y ; η) is complete and the sequence (fn )n≥1 is uniformly Cauchy then there
unif.
exists f : X → Y such that fn −−→ f .
n

11. Let ∅ =
6 X be a set, let (V, k · kV ) be a complete normed space and let (fn )n≥1
be a sequence of functions fn : X → V . Show that:
(1) If ∞
P
n=1 fn absolutely converges then it converges. P∞
(2) (Weierstraß Test) If kf n (x)k ≤ α n for all x ∈ X, n ∈ N, and n=1 αn < +∞
then ∞
P
n=1 f n converges absolutely and uniformly.
12. Let X 6= ∅ and let (V, k · k) be a normed space.
B(X; V ) := {f : X → V | f is a bounded function} .
For all f ∈ B(X; V ), let kf kX := supkf (x)k. Show that B(X; V ) is a vector space and
x∈X
k · kX is a norm, called the supremum norm on X.
13. Investigate the uniform convergence of the sequences
1 t t t
xn (t) = ( , ), yn (t) = ( , ), n ∈ N,
1 + nt n 1 + nt n
on [0, 1].
14. Given f : R2 → R2 defined by
f (x, y) = (x + 1, y − 1),
k·k
find kf kD := supD kf k2 , where D = B1 2 (0).

15. Consider the series ∞

X
xn y n .
n=0
k·k∞ k·k
Investigate its uniform convergence on: (a) B1 (0); (b) B1 2 (0).