ECOS2903 Lecture 2: Continuity and differentiation

Jiemai Wu∗
10 March 2021

1 Continuity

We all know what a continuous function looks like and what a discontinuous function looks
like, but can we articulate their signature features in general definition that works for multi-variate
functions whose domain and target space could be multidimensional subsets of Rn ?

Definition 1. (Page 293 of Simon-Blume)

Let f : Rk → Rm be a function. Let x0 be a vector in Rk , and y = f (x0 ) its image. Let {xn }∞
n=1
be any sequence in Rk . The function f is continuous at x0 if:
∞ m
{xn }∞
n=1 converges to x0 ⇒ the sequence { f (xn )}n=1 in R converges to f (x0 ) .

f is a continuous function if it is continuous at every point in its domain.

Example 1. Prove that f : R → R s.t. f (x) = 2x is a continuous function.

Proof. Let x0 be an arbitrary real number and let {xn }∞

n=1 be any sequence in R such that it con-
verges to x0 , i.e.,
lim xn = x0 .
n→∞
∗
If you spot an error, please contact me at [email protected]

1
 Because f (x) = 2x for all x ∈ R, f (xn ) = 2xn for all n. Therefore,

lim f (xn ) = lim 2xn = 2 × lim xn = 2x0 = f (x0 ) .

n→∞ n→∞ n→∞

This proves that f is continuous at x0 . Because the choice of x0 was arbitrary, f is continuous
on the entire domain R.

Example 2. Prove that f : R → R s.t.

(
1 when x > 0
f (x) =
−1 when x ≤ 0

is not a continuous function.

Proof. Proving that f is not continuous on its entire domain is equivalent to proving the following
statement (remember the swan exercise for quantifiers?)
There exists some x0 ∈ R such that there is a sequence {xn }∞n=1 in R that converges to x0 , but
∞
the sequence { f (xn )}n=1 does not converge to f (x0 ).
Let’s find such an example.
Let x0 = 0, so f (x0 ) = −1.
Let xn = n1 for n = 1, 2, 3, ... Then,

lim xn = 0 = x0 .
n→∞

However, because f (xn ) = 1 for all finite n,

lim f (xn ) = 1 6= f (x0 ) .

n→∞

This proves that f is not continuous at x0 = 0, so f is not a continuous function.

Fact 1. Let {xn }∞ ∞ m

n=1 and {yn }n=1 be convergent sequences of vectors in R with

lim xn = x and lim yn = y.

n→∞ n→∞

Let {cn }∞
n=1 be a convergent sequence of real numbers with

lim cn = c.
n→∞

Then, the sequence of vectors {cn xn + yn }∞

n=1 converges to

lim cn xn + yn = cx + y.
n→∞

(This is Theorem 12.6 in Section 12.2 of Simon-Blume.)

Theorem 1. Let f and g be functions from Rk to Rm . Suppose that f and g are continuous at v.
Then, f + g and f − g are both continuous at v.
(This is Theorem 13.4 of Simon-Blume.)

2
Proof. Let {vn }∞ n=1 be a converging sequence whose limit is v. By continuity, f (vn ) converges to
f (v) and g (vn ) converges to g(v).
Using the notations and results of Fact 1:
(1) Show that f + g is continuous.
If we let cn = 1 for all n, xn = f (vn ), and yn = g (vn ) then
lim ( f + g) (vn ) = lim [ f (vn ) + g (vn )] by definition of f + g
n→∞ n→∞
= lim f (vn ) + lim g (vn ) by Fact 1
n→∞ n→∞
= f (v) + g(v) by continuity of f and g
= ( f + g) (v) by definition of f + g
This proves that f + g is continuous at v.
(2) Show that f − g is continuous.
If we let cn = −1 for all n, xn = g (vn ), and yn = f (vn ) then
lim ( f − g) (vn ) = lim [ f (vn ) − g (vn )] by definition of f − g
n→∞ n→∞
= lim f (vn ) − lim g (vn ) by Fact 1
n→∞ n→∞
= f (v) − g(v) by continuity of f and g
= ( f − g) (v) by definition of f − g
This proves that f − g is continuous at v.

2 Differentiation
Is every continuous function differentiable? The answer is no. In other words, continuity is ne-
cessary but not sufficient for differentiability. Below is an example that is continuous but not
differentiable. In general, if a continuous function has a “kink”, then its slope is not defined at the
kink and, therefore, its derivative does not exist at the kink.

3
 0
Definition 2. A function is differentiable at x = a if it is continuous at a and its derivative f
0
exists at a. If its derivative f (x) is itself a continuous function, then we say that f is continuously
differentiable.

Notation: C0 , C1 , C2 , ......
A function f is C0 on domain D if it is continuous at every point of D.
0
A function f is C1 on domain D if its first derivative f exists and is continuous at every point
of D.
0 00
A function f is C2 on domain D if its first and second derivatives f , f exist and are con-
tinuous at every point of D.
0 00
A function f is Ck on domain D if its first k derivatives f , f , ..., f (k) all exist and are
continuous at every point of D.
A function f is C∞ or “smooth” on domain D if its derivative f (n) exits for all positive
integers n at every point of D.

• If a function is Ck+1 , then it must be Ck .

• The definition for C∞ implies that all derivatives f (n) are also continuous for all n.
Because differentiability ⇒ continuity, f (n) exist for all n ⇒ f (n−1) is continuous for
all n⇔ f (n) is continuous for all n.

Remark 1. In economics, we prefer to work with models whose functions are at least C2 .
0 00
For example, if f (x) = GDP in year x, then f represents the growth of the GDP, and f
represents the change in the growth of the GDP.

2.1 Group activity: Can you make your high school math teacher proud?
0
Let f be a well-defined function with a proper domain and target space. Evaluate f (x) at x = 2.
Your answer should be a number.

1.
f (x) = x7 9x2 − 1 + 4x3

2.
11x3 − 1
f (x) =
x
3.
2 −6x−6
f (x) = 6e6x

4.  
3x + 3
f (x) = ln
x3
Answer:
0
1. f (x) = 7x6 9x2 − 1 + x7 · 18x + 12x2 . This is equal to 20336 when x = 2.

4
 0 3
2. f (x) = 33x − 11xx2−1 . This is equal to 117
4 = 44.25 when x = 2.
0 6x2 −6x−6
3. f (x) = 6(12x − 6)e . This is equal to 108e6 ≈ 43570.3 when x = 2.
0 x3
· −3x−4 (3x + 3) + 3x−3 . This is equal to − 67 ≈ −1.16667 when x = 2.
 
4. f (x) = 3x+3
If you are struggling with some of these questions, please review the “math prerequisite”
page on Canvas: https://canvas.sydney.edu.au/courses/31039/pages/math-prerequisites
If you are still struggling, please post your question on Ed.

2.2 L’Hôpital’s rule

The derivative of a function is defined by a limit. Conversely, we can also use the rules of derivat-
ives to calculate limits that cannot be calculated directly. For example, we cannot directly calculate
the following limits because quotients of 0/0 or ∞/∞ are undefined.
Example 3. The following limits cannot be calculated by directly taking the ratio of the two limits.
ex − 1 0
lim = “ ”.
x→0 x 0
ln x ∞
lim = “ ”.
x→∞ x ∞
Fortunately, here is a very handy rule to help us.
L’Hôpital’s rule: Suppose that f and g are differentiable in an interval (α, β ) that contains a,
f (x) 0
except possibly a, and suppose that g(x) tends to 00 or ∞
∞
as x tends to a. If g (x) 6= 0 for all
0
f (x)
x 6= a in (α, β ), and if lim 0 = L, then
x→a g (x)
0
f (x) f (x)
lim = lim 0 = L.
x→a g(x) x→a g (x)

This is true whether L is finite, ∞, or −∞.

Let’s apply this rule to calculate the limits in the last example
Revisit last example. The following limits can be calculated using L’Hôpital’s rule.

ex − 1 ex
lim = lim = 1.
x→0 x x→0 1

ln x 1/x
lim = lim = 0.
x→∞ x x→∞ 1

3 Compact sets (next lecture)

This section builds the foundation for an important theorem that we will learn in Lecture 3: Ex-
treme Value Theorem (also called Weierstrass’s Theorem).

5
3.1 Bounded sets and compact sets: definitions
Definition 3. A set S in Rn is bounded if there exists a finite number B such that |x| ≤ B for all
x ∈ S.
In other words, a set S in Rn is bounded if S can be contained in some ball in Rn with a finite
radius.

 [0, 1] are2both bounded. (0, ∞) and

Example 4. (0, 1) and [0, ∞) are both+unbounded.
[0, 1]×(−1, 0) = (x, y) ∈ R | x ∈ [0, 1], y ∈ (−1, 0) is bounded. R = (x, y) ∈ R2 | x ≥ 0, y ≥ 0


is unbounded.

Definition 4. A set S in Rn is compact if it is both closed and bounded.

Example 5. [0, 1]2 = [0, 1] × [0, 1] is compact. (0, 1)2 = (0, 1) × (0, 1) is not compact because it is
not closed. [0, ∞) is not compact because it is not bounded.

3.2 Continuous function on a bounded/closed/compact domain

Let f : D → R be a real-valued (i.e., f (x) is a real number for all x in the domain) function whose
domain D is a subset of R. Can you construct examples such that:

1. D is bounded but not closed, f is continuous, and f does not have a finite “highest value” on
D.

2. D is closed but not bounded, f is continuous, and f does not have a finite “highest value” on
D.

3. D is closed and bounded, f is discontinuous, and f does not have a finite “highest value” on
D.

Here are possible examples.

1. D = (0, 1) and f (x) = x.

2. D = R and f (x) = x.

3. D = [−1, 1] and (
1−x when x ∈ (0, 1)
f (x) =
−1 when x ∈ [−1, 0]