Chapter 4: Continuity

Wissam Raji

Edited by Selena Nadjarian

Contents
1 Limits and Continuity in Metric Spaces 1

2 Continuous Functions on Compact Sets 6

3 Uniform Continuity and Discontinuities 9

4 Monotone Functions, Infinite Limits, and Inverse Mappings 15

1 Limits and Continuity in Metric Spaces

Definition 1.1. Let X and Y be metric spaces. Suppose,

f :E⊂X→Y

and let p ∈ E ′ . We say that

lim f (x) = q, q ∈ Y,
x→p

if ∀ ε > 0 there exists δ > 0 such that if x ∈ E and dX (x, p) < δ, then


dY f (x), q < ε.

Example 1.2. If X = Rk , Y = R, and f : E ⊂ Rk → R, then

lim f (x) = ℓ
x→p

means that ∀ ε > 0 there exists δ > 0 such that if x ∈ E and |x − p| < δ, then

|f (x) − ℓ| < ε.

1
 f (x)

f (x)

q+ϵ

q−ϵ

x
p−δpp+δ

Figure 1: The ϵ − δ definition of limits is visualized as such: for a given ϵ-band (red) around q,
there exists a δ-band (green) around p such that the graph of f (x) within the δ-band is entirely
contained within the ϵ-band.

Theorem 1.3 (Sequential criterion for limits). Let f : E ⊂ X → Y and p ∈ E ′ . Then,

lim f (x) = q
x→p

if and only if for every sequence {pn } ⊂ E with pn → p and pn ̸= p for all n, we have:

lim f (pn ) = q.
n→∞

In other words, a function approaches q at p if and only if every sequence approaching p makes
the function approach q.
Corollary 1.4. If f has a limit at p, then this limit is unique.
Definition 1.5. The function f : E ⊂ X → Y is said to be continuous at p ∈ E if ∀ ε > 0 there
exists δ > 0 such that


x ∈ E and dX (x, p) < δ =⇒ dY f (x), f (p) < ε.

Example 1.6. Let f : E ⊂ R → R where

E = (1, 2) ∪ {3}.

1 2 3

The line segment over (1, 2) has an open endpoint near x = 2, and an isolated point at x = 3. The
previous definition is true at p = 3, because we can choose δ > 0 such that

(3 − δ, 3 + δ) ∩ E = {3}.

Then,
|f (x) − f (3)| = |f (3) − f (3)| = 0 < ε.
Also, f is continuous at each point p ∈ (1, 2).

2
Remark 1.7. This example illustrates an important point: if p is an isolated point of the domain
E, then every function f : E → Y is automatically continuous at p. Indeed, since p is isolated, there
exists δ > 0 such that
Bδ (p) ∩ E = {p}.
Hence, whenever x ∈ E and d(x, p) < δ, we must have x = p, and therefore
dY (f (x), f (p)) = 0 < ε.
This shows that the ε–δ condition for continuity is automatically satisfied at isolated points.
This is a good reminder that continuity depends not only on the formula for f , but also on the
domain E, on which f is defined.
From the definitions of limit and continuity, if p ∈ E is a limit point, then
f is continuous at p ⇐⇒ lim f (x) = f (p).
x→p

Definition 1.8. The function f : E ⊂ X → Y is said to be continuous on E if it is continuous at

every point p ∈ E.
Example 1.9. Let X = R, Y = R, and
1
f (x) = sin , x ∈ R\{0}
x
We claim that
1
lim sin does not exist.
x→0 x
Proof: Consider the sequence
1
xn = , n = 1, 2, . . .
nπ
Then xn → 0 as n → ∞, and
f (xn ) = sin(nπ) = 0, =⇒ f (xn ) → 0.
On the other hand, consider the sequence
1
yn = , n = 1, 2, . . .
π/2 + 2nπ
Then yn → 0 as n → ∞, and
π 
f (yn ) = sin + 2nπ = 1, =⇒ f (yn ) → 1
2
Since the limits along these two sequences are different, the sequential criterion for limits implies
that
1
lim sin does not exist.
x→0 x
Theorem 1.10 (Squeeze theorem). Let A ⊂ R and let f, g, h : A → R. Let c ∈ A′ . If
f (x) ≤ g(x) ≤ h(x), ∀x ∈ A, x ̸= c,
and
lim f (x) = L = lim h(x),
x→c x→c
then
lim g(x) = L.
x→c

3
Using the Sequential Criterion. Let {xn } ∈ A be any sequence such that xn → c, with xn ̸= c ∀n.
Then the inequality holds along the sequence:

f (xn ) ≤ g(xn ) ≤ h(xn ), ∀n

Taking the lim sup as n → ∞ we get:

lim sup f (xn ) ≤ lim sup g(xn ) ≤ lim sup h(xn ).

n→∞ n→∞ n→∞

Since the outer limits are both L, we obtain

L ≤ lim sup g(xn ) ≤ L, =⇒ lim sup g(xn ) = L

n→∞ n→∞

Similarly, taking lim inf in the inequality gives

lim inf g(xn ) = L.

n→∞

Since,
lim inf g(xn ) = lim sup g(xn ) = L,
n→∞ n→∞
It follows that
limn→∞ g(xn ) = L.
Then, by the sequential criterion for limits, this implies:

lim g(x) = L.
x→c

Example 1.11. Show that

1
lim x sin = 0.
x→0 x
We observe that ∀x ̸= 0
1
−|x| ≤ x sin ≤ |x|.
x
Since limx→0 −|x| = limx→0 |x| = 0, the squeeze theorem implies that:
1
lim x sin = 0.
x→0 x
Theorem 1.12 (Algebra of limits). Let f : E ⊂ X → R and g : E ⊂ X → R. If

lim f (x) = A, lim g(x) = B,

x→p x→p

then:
1. lim (f + g)(x) = A + B,
x→p

2. lim (f g)(x) = AB,

x→p
 
f A
3. lim (x) = , provided B ̸= 0.
x→p g B

4
Theorem 1.13. Let f and g be continuous on a metric space X. Then f + g and f g are continuous
on X, and f /g is continuous on every point x ∈ X such that g(x) ̸= 0.
Theorem 1.14 (Composition of continuous functions). Let X, Y, and Z be metric spaces, let
f : E ⊂ X → Y , and g : f (E) → Z. Define

h(x) = g(f (x)), x ∈ E.

If f is continuous at p ∈ E and g is continuous at f (p), then h is continuous at p.

Proof. Let ε > 0. Since g is continuous at f (p), there exists η > 0 such that


y ∈ f (E) and dY (y, f (p)) < η =⇒ dZ g(y), g(f (p)) < ε.

Since f is continuous at p, there exists δ > 0 such that

x ∈ E and dX (x, p) < δ =⇒ dY f (x), f (p) < η.

Combining the two implications, ∀x ∈ E with dX (x, p) < δ, we have

dZ h(x), h(p) = dZ g(f (x)), g(f (p)) < ε.

Hence, h is continuous at p.

Theorem 1.15 (Open-set characterization). Let X and Y be metric spaces. A function f : X → Y

is continuous on X if and only if

f −1 (V ) is open in Xfor every open set V ⊂ Y

Proof. (⇒) Suppose f is continuous on X. Let V ⊂ Y be an open set. We show that f −1 (V ) is

open in X. Let p ∈ f −1 (V ). Then f (p) ∈ V , and since V is open there exists ε > 0 such that


Nε f (p) = {y ∈ Y : dY (y, f (p)) < ε} ⊂ V.

By continuity of f at p, there exists δ > 0 such that

dX (x, p) < δ =⇒ dY f (x), f (p) < ε.

Hence ∀x ∈ Nδ (p), we have f (x) ∈ V , i.e.

Nδ (p) ⊂ f −1 (V ).

Therefore, f −1 (V ) is open in X.
(⇐) Suppose now that f −1 (V ) is open in X for every open set V ⊂ Y . Fix p ∈ X and ε > 0.
Define

V = Nε f (p) := {y ∈ Y : dY (y, f (p)) < ε}.
Then V is open in Y , so f −1 (V ) is open in X and contains p. Hence, there exists δ > 0 such that

Nδ (p) ⊂ f −1 (V ).

Therefore, if dX (x, p) < δ, then x ∈ f −1 (V ), which means

f (x) ∈ V =⇒ dY f (x), f (p) < ε

Thus, f is continuous at p.

5
Corollary 1.16. A function f : X → Y is continuous if and only if f −1 (C) is closed in X for
every closed set C ⊂ Y .

Proof. Recall that a set E ⊂ Y is closed if and only if its complement E c is open. Moreover, inverse
images preserve complements:
f −1 (E c ) = (f −1 (E))c
Thus, f −1 (C) is closed in X if and only iff f −1 (C c ) is open in X. By Theorem 1.15, this is equivalent
to f being continuous.

2 Continuous Functions on Compact Sets

Theorem 2.1. Let f : X → Y be continuous, where X is a compact metric space. Then f (X) is a
compact subset of Y .

Proof. Let {Gα } be an open cover of f (X), i.e.

[
f (X) ⊂ Gα
α

where each Gα is open in Y . Since f is continuous, the preimage of each Gα is open in X, and we
have [
X= f −1 (Gα ).
α

Because X is compact, there exists a finite subcover:

X ⊂ f −1 (Gα1 ) ∪ · · · ∪ f −1 (Gαn ). (1)

Applying f to both sides and using the fact that

f (f −1 (E)) ⊂ E

for any set E, we obtain

f (X) ⊂ Gα1 ∪ · · · ∪ Gαn . (2)
Thus, {Gα1 , ..., Gαn } is a finite subcover of f (X), and hence f (X) is compact

Remark 2.2. One must be careful not to overgeneralize this theorem.

A continuous image of a compact set is compact, but this conclusion may fail if the domain is
only bounded or only closed.
For example, the set (0, 1) is bounded but not compact. The function
1
f : (0, 1) → R, f (x) = ,
x
is continuous but its image (1, ∞) is not bounded.
Likewise, a continuous image of a closed set need not be closed in general. Thus, compactness is
the key hypothesis here.

Remark 2.3. Let E ⊂ X be compact and f : E → Y be continuous, then f (E) is compact.

6
Note. For any set E, we always have

E ⊂ f −1 (f (E)).

The proof is left to the reader in the notes.

Theorem 2.4. Let

f⃗ : X → Rk
be continuous where X is compact. Then f⃗(X) is closed and bounded. In particular, f⃗ is bounded.

Proof. By the previous theorem, the image, f⃗(X), is compact in Rk . By the Heine–Borel theorem,
every compact subset of Rk is closed and bounded. Therefore, f⃗(X) is closed and bounded, so in
particular it is bounded.

Definition 2.5. A map

f⃗ : E ⊂ X → Rk
is said to be bounded if there exists M > 0 such that

|f⃗(x)| ≤ M, ∀x ∈ E.

Theorem 2.6 (Extreme value theorem). Let f : X → R be continuous, where X is compact. Define

M = sup f (p), m = inf f (p).

p∈X p∈X

Then, there exist points p, q ∈ X such that

f (p) = M, f (q) = m.

Remark 2.7. The theorem states that a continuous function defined on a compact set attains both
its maximum and minimum values at points in the domain.

Example 2.8. Consider the function

f (x) = x2 , f : [−1, 2] → R,

This function is continuous on a compact set. We have

M = 4, m = 0,

and these are attained at

f (2) = 4, f (0) = 0.

Example 2.9. Consider the function

f : R → R, f (x) = ex .

Then f is continuous, but it does not attain a maximum on R, since it is unbounded above. Its
infimum is
m=0
but this value is not attacined. Hence, the supremum and infimum are both not attained ∈ R.

7
Remark 2.10. The Extreme Value Theorem guarantees attainment of maximum and minimum
values only when the domain is compact. Without compactness, even very well-behaved continuous
functions may fail to attain their extreme values.

Proof of the extreme value theorem. Since f (X) is closed and X is compact, the image f (X) is
compact in R. Hence, f (X) is closed and bounded.
Therefore,
sup f (X), inf f (X) ∈ f (X).
It follows that there exist points p1 , p2 ∈ X such that

M = f (p1 ) = sup f (X), m = f (p2 ) = inf f (X).

Recall the auxiliary facts stated in the notes:

1. If E ⊂ R is closed and bounded and α = sup E, then α ∈ E.

2. E is closed if and only if E = E.

Theorem 2.11. Let f : X → Y be continuous. If E ⊂ X is connected, then f (E) is connected.

Proof. We proceed by contradiction. Suppose f (E) is not connected. Then there exists nonempty,
disjoint, separated sets A, B ⊂ Y such that

f (E) = A ∪ B, A, B ̸= ∅, A ∩ B = ∅, B ∩ A = ∅.

Define
G = E ∩ f −1 (A), H = E ∩ f −1 (B).
Then
E = G ∪ H, G, H ̸= ∅.
Moreover, since f is continuous and A, B are separated in F (E), their preimage f −1 (A) and
f −1 (B)are separated in X. Intersecting with E, it follows that G and H are separated in E. In
particular

G ∩ H = ∅, G ∩ H = ∅, G∩H =∅
Thus, E can be written as the union of two nonempty, disjoint, separated sets G and H, which
means that E is disconnected. This contradicts the assumption that E is connected.
Therefore, f (E) must be connected.

Theorem 2.12 (Intermediate value theorem). Let f : [a, b] → R be continuous. If

f (a) < c < f (b),

then there exists x ∈ (a, b) such that

f (x) = c.

Note. This theorem states that a continuous real function assumes all intermediate values on an
interval.
Note. A similar result holds if f (a) > f (b).

8
Proof. The interval [a, b] is connected. So, by the previous theorem, image f ([a, b]) is connected in
R. By the characterization of connected sets in R, if E ∈ R is connected then for any x, y ∈ E with
x < z < y, we have z ∈ E.
When we apply this to:

E = f ([a, b]) = {f (x) : a ≤ x ≤ b}, x = f (a), y = f (b), z = c

Sine c ∈ (f (a), f (b)) , there exists x ∈ (a, b) such that f (x) = c

Remark 2.13. Assume f (a) < c < f (b) and define

E = {x ∈ [a, b] : f (x) < c}.

Then E is nonempty and bounded above, so x0 = sup E exists. By continuity of f at x0 , neither

f (x0 ) < c nor f (x0 ) > c is possible. Hence

f (x0 ) = c.

This classical proof highlights how continuity interacts with the completeness of R

3 Uniform Continuity and Discontinuities

Definition 3.1. Let f : X → Y . We say that f is uniformly continuous on X if for every ε > 0
there exists δ > 0 such that


p, q ∈ X, dX (p, q) < δ =⇒ dY f (p), f (q) < ε.

Remark 3.2 (Continuity vs. Uniform Continuity). The conceptual difference between continuity
and uniform continuity is ”who chooses δ first.”

• In standard continuity, δ depends on both ϵ and the specific point p, written as δ = δ(ϵ, p).
As p moves around the domain, the required δ might shrink toward zero (e.g., when a graph
becomes steeper).

• In uniform continuity, δ depends only on ϵ, written as δ = δ(ϵ). A single, ”universal” δ works

for the entire domain simultaneously. Geometrically, this means we can slide a rigid rectangle
of height 2ϵ and width 2δ along the entire graph, and the curve will always exit through the
sides of the rectangle, never poking out of the top or bottom edges.

Note. Clearly, if f is uniformly continuous, then f is continuous. The converse is not necessarily
true, as the following examples shows.

Example 3.3. Consider the function

1
f (x) = , x ∈ (0, 1)
x
First, f is continuous at every point in (0, 1) because it is a rational function whose denominator
is never zero on this domain.
However, f is not uniformly continuous on (0, 1). Suppose f were uniformly continuous. Then
for ever ϵ > 0 there must exists a single δ > 0 such that for all x, y ∈ (0, 1) satisfying |x − y| < δ we
have |f (x) − f (y)| < ϵ

9
 Let ϵ = 1 and define 2 sequences approaching 0:
1 1
xn =
, yn = n∈N
n n+1
The horizontal distance between these points is:

1 1 1
|xn − yn | = − = .
n n+1 n(n + 1)
By choosing n large enough, |xn − yn | can be made smaller than any proposed δ.
However, the vertical distance is:

|f (xn ) − f (yn )| = |n − (n + 1)| = | − 1| = 1.

which does not satisfy |f (xn ) − f (yn )| < ϵ = 1.

as x → 0, the function becomes infinitely steep. The required δ depends on x and shrinks to
zero. Therefore, no single δ works for the entire interval, so f us not uniformly continuous.

f (x)

ϵ
1
f (x) = x
x
δ1
δ2

Figure 2: For f (x) = 1/x, maintaining the same vertical error bound (ϵ) requires a much smaller
horizontal neighborhood (δ2 ≪ δ1 ) as x → 0. Thus, a single uniform δ cannot be chosen for the
entire interval.

Example 3.4. A noncompact domain does not automatically prevent uniform continuity.
For instance, consider:
f : R → R, f (x) = x,
Given ε > 0, choose δ = ε. Then for all x, y ∈ R

|x − y| < δ =⇒ |f (x) − f (y)| = |x − y| < ε.

Hence, f is uniformly continuous on R
Another example:
f : R → R, f (x) = arctan x.

10
Since
| arctan x − arctan y| ≤ |x − y|, ∀x, y ∈ R
f is uniformly continuous on R as well.

Remark 3.5. Compactness guarantees uniform continuity, but uniform continuity may still hold
on noncompact domains, as seen in the above examples.

Theorem 3.6. Let f : X → Y be continuous, and suppose X is a compact metric space. Then f is
uniformly continuous on X.

Proof. Let ε > 0. Since f is continuous at each p ∈ X, there exists η(p) > 0 such that

ε
q ∈ X, dX (p, q) < η(p) =⇒ dY f (p), f (q) < . (*)
2
Define  
η(p)
J(p) = q ∈ X : dX (p, q) < = Nη(p)/2 (p).
2
Each J(p) is open, and the family {J(p) : p ∈ X} forma an open cover of X. By compactness, there
exists a finite subcover
J(p1 ) ∪ · · · ∪ J(pn ) = X.
Set
1 
δ := min η(p1 ), . . . , η(pn ) .
2
We claim that

dX (p, q) < δ =⇒ dY f (p), f (q) < ε. (**)
Indeed, let dX (p, q) < δ. Then p ∈ J(pm ), for some m, hence
1
dX (p, pm ) < η(pm ).
2
Moreover,
1
dX (q, pm ) ≤ dX (q, p) + dX (p, pm ) < δ + η(pm ) ≤ η(pm ).
2
By (*) we obtain



ε ε
dY f (p), f (q) ≤ dY f (p), f (pm ) + dY f (pm ), f (q) < + = ε.
2 2
This proves (**) and hence f is uniformly continuous on X.

Example 3.7. Let

1
f : E ⊂ R → R, f (x) = , E = (x0 − 1, x0 + 1) \ {x0 }.
x − x0
Then E is not compact, and f is continuous on E, but f is not uniformly continuous on E.

11
Proof. Suppose for contradiction that f is uniformly continuous. Then for some ε > 0 there exists
δ > 0 such that
|t − x| < δ, t, x ∈ E =⇒ |f (t) − f (x)| < ε. (1)
Choose N1 , N2 > 0 such that

1 1
− < δ, N2 − N1 > ε.
N2 N1

and set
1 1
t = x0 + , x = x0 + .
N1 N2
Then
1 1
|x − t| = − < δ,
N2 N1
but
1 1
|f (x) − f (t)| = − = |N2 − N1 | > ε,
x0 + 1/N2 − x0 x0 + 1/N1 − x0
contradicting (1). Therefore, f is not uniformly continuous on E.

Example 3.8. Let

1
g : E ⊂ R → R, E = (x0 − 1, x0 + 1) \ {x0 }, g(x) = .
1 + (x − x0 )2

Then g is continuous on E and E is not compact. Moreover,

1
< g(x) < 1, x ∈ E,
2
so g is bounded and
sup g(x) = 1.
x∈E

But g has no maximum on E.

Example 3.9. Let

x2
h : R → R, h(x) = .
1 + x2
Then h is continuous on R and bounded:

0 ≤ h(x) < 1, x ∈ R.

Moreover,
sup h(x) = 1,
x∈R

but since R is not compact, h does not attacn a maximum in R. Note also that
1
h(x) = 1 − → 1 as x → ∞
1 + x2

12
 f (x)

x
x1 x2 x3

Figure 3: Classification of discontinuities: a removable discontinuity at x1 (first kind), a jump

discontinuity at x2 (first kind), and an infinite/essential discontinuity at x3 (second kind).

Example 3.10. Let

E = {±1, ±2, ±3, . . . }.
Then E is unbounded, but any function f : E → R is uniformly continuous. Indeed, given ε > 0,
choose δ = 1/2. Then if x, y ∈ E and |x − y| < δ, necessarily x = y, so

|f (x) − f (y)| = 0 < ε.

Definition 3.11. Let f : E ⊂ R → R. If x ∈ E and f is discontinuous at x, we say that f has a

discontinuity at x.

Definition 3.12. Let f be defined on (a, b).

• If a < x ≤ b, we write
f (x+) = q
if f (tn ) → q for all sequences {tn } in (x, b) such that tn → x.

• If a ≤ x < b, we write
f (x−) = q
if f (tn ) → q for all sequences {tn } in (a, x) such that tn → x.

Clearly,
lim f (t) = α ⇐⇒ f (x+) = f (x−) = α.
t→x

Definition 3.13. Let f : (a, b) → R. If f (x+ ) and f (x− ) exist but f is discontinuous at x, then
we say that f has a discontinuity of the first kind, or a simple discontinuity, at x. Otherwise, the
discontinuity is said to be of the second kind.

Example 3.14 (Thomae’s Function). Consider the function

(
1
if x = pq is rational in lowest terms (with q > 0),
f : (0, 1) → R, f (x) = q
0 if x is irrational.

13
This function, often called the ”popcorn function,” has a fascinating and counterintuitive property.
It is discontinuous at every rational number, but continuous at every irrational number. This
demonstrates that the set of discontinuities of a function can be dense in the domain without the
function being nowhere continuous.

Note. From the previous definition, a simple discontinuity at x means either

1. f (x+ ) ̸= f (x− ), or

2. f (x+ ) = f (x− ) ̸= f (x).

Remark 3.15. The continuity at irrational points can be understood intuitively as follows:
If x0 is irrational, then near x0 there may be many rational numbers, but only finitely many of
them can have a small denominator. Since the value of the function at a rational number pq is 1q ,
the only rationals that lie very close to x0 must typically have a large denominator, and hence their
function values are very small.
Thus, near an irrational point, the values of the function are forced to stay close to 0, which
explains why the function is continuous there.

Example 3.16. Let (

1, 0 < x < 1,
f (x) =
0, 1 ≤ x < 2.
Then f has a simple discontinuity at x = 1.

Example 3.17. Let (

1, x rational,
f (x) =
0, x irrational.
Then f has a discontinuity of the second kind at every point x, because f (x+ ) and f (x− ) do not
exist. Indeed, every interval (x − δ, x + δ) contains both rational and irrational numbers.

Example 3.18. Let (

x, x rational,
f (x) =
0, x irrational.
At x = 0:
lim f (x) = 0 = f (0).
x→0

Indeed, given ε > 0, choose δ = ε. If |y| < δ, then:

• if y is irrational, then |f (y) − f (0)| = |0 − 0| < ε;

• if y is rational, then |f (y) − f (0)| = |y| < ε.

Hence f is continuous at x = 0.

At x ̸= 0: f (x+ ) and f (x− ) do not exist, because any interval (x − δ, x + δ) contains both rational
and irrational numbers. One may choose sequences of irrational numbers tn → x and rational
numbers sn → x such that {f (tn )} and {f (sn )} do not converge to any common limit.

Therefore f has a discontinuity of the second kind at every x ̸= 0.

14
Example 3.19. Let (
sin x1 , x ̸= 0,
f (x) =
0, x = 0.
We showed earlier that limx→0 f (x) does not exist. Hence f has a discontinuity of the second kind
at x = 0 because f (0+ ) and f (0− ) do not exist.

Nevertheless, f is continuous at every point x ̸= 0, since it is the composition of the two continuous
functions
1
x 7→ , x 7→ sin x.
x

4 Monotone Functions, Infinite Limits, and Inverse Mappings

Definition 4.1. Let f : (a, b) → R.

• f is monotonically increasing on (a, b) if

a < x < y < b =⇒ f (x) ≤ f (y).

• f is monotonically decreasing on (a, b) if

a < x < y < b =⇒ f (x) ≥ f (y).

Theorem 4.2. Let f : (a, b) → R be monotonically increasing on (a, b). Then f (x+ ) and f (x− )
exist at every point x ∈ (a, b). More precisely,

sup f (t) = f (x− ) ≤ f (x) ≤ f (x+ ) = inf f (t).

a<t<x x<t<b

Moreover, if a < x < y < b, then

f (x+ ) ≤ f (y )
A similar result holds for monotonically decreasing functions.

Proof. Let
A = sup f (t).
a<t<x

Since f (t) ≤ f (x) for all t ∈ (a, x), f (x) is an upper bound of the set {f (t) : a < t < x}. Hence, by
the definition of the supremum,
A ≤ f (x).
We show that A = f (x− ). Let ε > 0. By definition of A, there exists y such that

A − ε < y ≤ A.

Write y = f (z) for some z ∈ (a, x). Since f is increasing,

A − ε < f (x − δ) ≤ A. (1)

Since f is increasing,
f (x − δ) ≤ f (t) ≤ A, x − δ < t < x (2)

15
 From (1) and (2) we obtain,

A − ε < f (x − δ) ≤ f (t) ≤ A, x − δ < t < x,

so
|f (t) − A| < ε, x − δ < t < x.
Thus f (x− ) = A.

We have proved

sup f (t) = f (x− ) ≤ f (x). (3)

a<t<x

In the same way, one obtains

f (x) ≤ f (x+ ) = inf f (t). (4)
x<t<b

From (3) and (4), we conclude

sup f (t) = f (x− ) ≤ f (x) ≤ f (x+ ) = inf f (t).

a<t<x x<t<b

For the second part, fix x < y in (a, b). Apply the first part on (a, y) instead of (a, b) to get

f (x+ ) ≤ inf f (t), (5)

x<t<y

and on (x, b) instead of (a, b) to get

f (y − ) = sup f (t). (6)

x<t<y

From (5) and (6), we conclude

f (x+ ) ≤ f (y − ).

Corollary 4.3. Monotone functions do not have discontinuities of the second kind, since f (x+ )
and f (x− ) always exist.
Theorem 4.4. Let f : (a, b) → R be monotonic. Then the set of points of (a, b) at which f is
discontinuous is at most countable.

Proof. Without loss of generality, assume that f is increasing. Define

E = {x ∈ (a, b) : f is discontinuous at x}.

If x ∈ E, then there exists a rational number r(x) such that

f (x− ) < r(x) < f (x+ ).

Indeed, from the previous theorem,

f (x− ) ≤ f (x) ≤ f (x+ ),

and if f (x− ) = f (x+ ), then f would be continuous at x. Hence f (x− ) < f (x+ ) and therefore the
interval (f (x− ), f (x+ )) contains a rational number.

16
 If x1 ̸= x2 , say x1 < x2 , then

f (x− + − +
1 ) < r(x1 ) < f (x1 ) ≤ f (x2 ) < r(x2 ) < f (x2 ),

so in particular r(x1 ) ̸= r(x2 ). Therefore, the map

x 7→ r(x)

embeds E injectively into Q. Since Q is countable, E is at most countable.

f (x)

f (x0 +)
f (x0 ) Jump Gap
f (x0 −)

x
x0

Figure 4: For a monotonically increasing function, the left limit f (x0 −) and right limit f (x0 +)
always exist. The true function value f (x0 ) must lie between them, or coincide with one of them.
The size of the discontinuity is exactly f (x0 +) − f (x0 −).

Remark 4.5. The previous theorem gives a very useful qualitative picture of monotone functions:

• a monotone function always has one-sided limits;

• Therefore, it can only have jump discontinuities;

• The set of such discontinuities is at most countable.

Thus, although a monotone function need not be continuous, its discontinuities are highly
restricted.

Infinite limits and limits at infinity

Recall the finite-valued limit definition: if X, Y are metric spaces and f : E ⊂ X → Y with p ∈ E ′ ,
then
lim f (x) = q ∈ Y
x→p

means that for every ε > 0 there exists δ > 0 such that


x ∈ E, dX (x, p) < δ =⇒ dY f (x), q < ε.

17
 For the case f : E ⊂ R → R, this becomes
lim f (x) = ℓ
x→p

if for every ε > 0 there exists δ > 0 such that

|x − p| < δ, x ∈ E =⇒ |f (x) − ℓ| < ε. (1)
Sometimes we need to take limits when p ∈ {−∞, +∞} and/or ℓ ∈ {−∞, +∞}. We therefore
consider the extended real system
{−∞} ∪ R ∪ {+∞}.
For example:
• f (x) → ∞ as x → a, a ∈ R, means:
∀M ∈ R, ∃δ > 0 such that |x − a| < δ, x ∈ E =⇒ f (x) > M.

• f (x) → a as x → ∞, a ∈ R, means:
∀ε > 0, ∃M ∈ R such that for all x ∈ E, x > M =⇒ |f (x) − a| < ε.

• f (x) → ∞ as x → ∞ means:
∀M ∈ R, ∃N ∈ R such that for all x ∈ E, x > N =⇒ f (x) > M.

With these notions of limit, the usual algebraic rules still hold.
Theorem 4.6. Let f, g : E ⊂ R → R. Suppose
f (t) → A, g(t) → B, t → x,
where A and B belong to
{−∞} ∪ R ∪ {+∞}.
Then, whenever the right-hand side is defined,
1. f (t) → A′ =⇒ A′ = A,
2. (f + g)(t) → A + B,
3. (f g)(t) → AB,
4. (f /g)(t) → A/B.
The right members in (2), (3), and (4) must be defined; for example, the forms
∞ A
∞ − ∞, 0 · ∞, ,
∞ 0
are not defined in this context.
Example 4.7. Let
x2 1
f (x) = 2
=
1+x 1 + 1/x2
1 1
Since x2
→ 0 as x → ∞, we have 1 + x2
→ 1 and thus,

x2
→1 (x → ∞).
1 + x2

18
Inverse mappings
Theorem 4.8. Suppose f is a continuous one-to-one mapping of a compact metric space X onto a
metric space Y . Then the inverse mapping
f −1 : Y → X, f −1 (f (x)) = x, x ∈ X,
is continuous.

Proof. Let V ⊂ X be open, and let g = f −1 . We want to show that g −1 (V ) is open in Y . But
g −1 (V ) = {y ∈ Y : g(y) ∈ V }.
Since f (V ) = g −1 (V ), it is enough to show that f (V ) is open.
V c is closed in X, Since X is compact, V c is compact. f is continuous, so
f (V c ) is compact and
c
therefore closed. Finally, because f is one-to-one and onto, f (V c ) = f (V )
Hence f (V ) is open, so g is continuous.

Example 4.9. Consider the function

f : [0, 1] → [0, 1], f (x) = x3 .
This function is continuous and one-to-one on the compact set [0, 1], mapping onto [0, 1].
Its inverse f −1 (y) = y 1/3 , is continuous by the theorem.
On compact domains, a continuous bijection always has a continuous inverse. Without compact-
ness, this may fail.
Theorem 4.10 (Continuity preserves sequential convergence). If f : E ⊂ X → Y is continuous,
and if {xn } ⊂ E with xn → x ∈ E, then
f (xn ) → f (x).
Proof. Let ε > 0. Since f is continuous at x, there exists δ > 0 such that


dX (y, x) < δ, y ∈ E =⇒ dY f (y), f (x) < ε. (1)
Since xn → x, there exists N such that
dX (xn , x) < δ, ∀n ≥ N. (2)
From (1) and (2),

dY f (xn ), f (x) < ε, ∀n ≥ N.
Hence f (xn ) → f (x).

Example 4.11. Use the continuity of the exponential function and the convergence
ln n
→0 (n → ∞)
n
to show that √
n
n → 1.
Let
√
n
ln n
yn = ln n= → 0.
n
Consider the continuous function f (x) = ex . Then, f (yn ) → f (0) = 1, i.e.,
√n
ln(n)
n=e n →1

19