Differential Calculus – One Variable

Rolle’s Theorem:
If a function 𝑓 is

a) continuous in [a, b]
b) differentiable in (a, b)
c) 𝑓(𝑎) = 𝑓(𝑏)
Then ∃ 𝑐 ∈ (𝑎, 𝑏) s.t. 𝑓 ′ (𝑐 ) = 0

a b a b

Proof: Suppose M & m are maximum and minimum of 𝑓(𝑥) in [𝑎, 𝑏].
(It will always exist because of Weierstrass extreme value theorem as 𝑓 is
continuous in [a, b])

Case I: if 𝑀 = 𝑚 i.e. 𝑓(𝑥) = 𝑀 = 𝑚 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

This implies 𝑓 ′ (𝑥) = 0 ∀𝑥 ∈ (𝑎, 𝑏)

Case II: 𝑀 ≠ 𝑚. Then at least one of them must be different from

equal values of 𝑓(𝑎) and 𝑓(𝑏).

c
Minimum is different from f(a) and f(b) Maximum is different from f(a) and f(b) Both are different from f(a) and f(b)
1
 Differential Calculus – One Variable

Let 𝑀 = 𝑓(𝑐) be different. Since 𝑓 is differentiable in (𝑎, 𝑏),

𝑓 ′ (𝑐 ) exists. Note that 𝑓(𝑐) is the maximum value, then

𝑓 (𝑐 + Δ 𝑥) − 𝑓 (𝑐 ) ≤ 0 for Δ𝑥 > 0 or Δ𝑥 < 0

This implies:

𝑓(𝑐 + Δ𝑥 ) − 𝑓(𝑐 )
≤ 0, for Δ𝑥 > 0
Δ𝑥
and
𝑓(𝑐 + Δ𝑥 ) − 𝑓(𝑐 )
≥ 0, for Δ𝑥 < 0
Δ𝑥
Since 𝑓′(𝑐) exists, passing limit as Δ𝑥 → 0, we get
𝑓 (𝑐 + Δ𝑥) − 𝑓(𝑐 )
Lim = 𝑓′(𝑐) ≤ 0
Δ𝑥→0 Δ𝑥 (1)
(Δ𝑥>0 )
and
𝑓(𝑐 + Δ𝑥) − 𝑓(𝑐 )
Lim = 𝑓 ′ (𝑐 ) ≥ 0
Δ𝑥→0 Δ𝑥 (2)
(Δ𝑥<0 )

Inequality (1) and (2) implies 𝑓’(𝑐) = 0.

Remark 1: The conclusion of Rolle’s theorem may not hold for a

function that does not satisfy any of its conditions.

Ex 1: Consider
𝑥, 𝑥 ∈ [0, 1]
𝑓 (𝑥 ) = {
2 − 𝑥, 𝑥 ∈ (1, 2]
2
 Differential Calculus – One Variable

0 1 2

Note that f ′ (x) ≠ 0 for any x ∈ (1, 2). However, this does not
contradict Rolle's Theorem, since f′(1) does not exist.

Remark 2: The continuity condition for the function on the closed

interval [a, b] is essential.

Ex: Consider
𝑥, 𝑥 ∈ [0, 1)
𝑓 (𝑥 ) = {
0, 𝑥=1

0 1

Then, f is continuous and differentiable on (0, 1), and also

𝑓(0) = 𝑓(1). But f ′ (x) ≠ 0 for any x ∈ (0, 1).

Remark 3: The hypotheses of Rolle’s theorem are sufficient but not

necessary for the conclusion. Meaning, if all three hypotheses are
met then conclusion is guaranteed. Not necessary means if the
hypotheses are not met then you may (or may not) reach the
conclusion.
3
 Differential Calculus – One Variable

Example: Discuss the applicability of Rolle’s theorem to the function

𝑥2 + 1 , 𝑥 ∈ [0, 1]
𝑓 (𝑥 ) = {
3 − 𝑥, 𝑥 ∈ (1, 2]
Solution:
1) Continuity
𝑓 (1 + 0) = 𝑙𝑖𝑚 3 − (1 + 𝛥𝑥) = 𝑙𝑖𝑚 [2 − 𝛥𝑥] = 2 = 𝑓(1)
𝛥𝑥→0 𝛥𝑥→0
𝛥𝑥>0 𝛥𝑥 >0
2) Differentiability

𝑓 (1 + Δ𝑥 ) − 𝑓(1) (2 − Δ𝑥) − 2
𝑓 ′ (1 + 0) = lim = lim = −1
Δ𝑥→0 Δ𝑥 Δ𝑥→0 Δ𝑥
Δ𝑥>0 Δ𝑥>0

′(
𝑓 (1 + Δ𝑥 ) − 𝑓(1) (1 + Δ𝑥 )2 + 1 − 2
𝑓 1 − 0) = lim = lim
Δ𝑥→0 Δ𝑥 Δ𝑥→0 Δ𝑥
Δ𝑥<0 Δ𝑥<0
2
2Δ𝑥 + Δ𝑥
= lim =2
Δ𝑥→0 Δ𝑥
Δ𝑥<0
Thus 𝑓 ′ (1 + 0) ≠ 𝑓 1 − 0). This implies 𝑓 is not differentiable.
′(

Example: Using Rolle ’s Theorem, show that the equation

𝑥 13 + 7𝑥 3 − 5 = 0 has exactly one real root in [0, 1].

Solution: Let 𝑓(𝑥) = 𝑥 13 + 7𝑥 3 − 5 has two real roots, say 𝛼 and 𝛽

in [0, 1]. That is, we have 𝑓(𝛼) = 𝑓(𝛽 ) = 0. All hypotheses of
Rolle’s theorem are satisfied in [𝛼, 𝛽 ].
Rolle ’s Theorem implies 𝑓 ′ (𝑐 ) = 0 for some 𝑐 ∈ (𝛼, 𝛽 ).
⇒ 13 𝑐12 + 21𝑐 2 = 0 for some 𝑐 ∈ (𝛼, 𝛽). Note that 𝑐 > 0 as
𝛼 ≥ 0. It contradicts our assumption of two real roots.
On the other hand 𝑓(0) = −5 and 𝑓(1) = 3. It confirms the
existence of at least one root. Hence the function has exactly one
root.

4
 Differential Calculus – One Variable

Lagrange’s mean value theorem:

If a function 𝑓 is

a) continuous in [a, b]
b) differentiable in (a, b)

then there exists at least one value 𝑐 ∈ (𝑎, 𝑏) such that

𝑓 ( 𝑏 ) − 𝑓 (𝑎 )
= 𝑓′(𝑐)
𝑏−𝑎
In other words, there is at least one tangent line in the interval that
is parallel to the secant line that goes through the endpoints of the
interval.

𝑓(𝑏) − 𝑓(𝑎)

𝛼
𝑓(𝑏) − 𝑓(𝑎)
tan 𝛼 =
𝑏−𝑎
a b

Proof: Define a function

𝑓 (𝑏 ) − 𝑓 (𝑎 )
𝜙 (𝑥 ) = 𝑓 (𝑥 ) − [ ]𝑥
𝑏−𝑎

Note that the function 𝜙(𝑥) satisfies all the conditions of Rolle ’s
Theorem as 𝜙(𝑎) = 𝜙(𝑏), and continuity and differentiability
follows from the continuity and differentiability of 𝑓(𝑥). Rolle ’s
Theorem gives
𝑓(𝑏)−𝑓(𝑎)
𝜙 ′ (𝑐 ) = 0 for some 𝑐 ∈ (𝑎, 𝑏) ⇒ 𝑓 ′ (𝑐 ) − = 0.
𝑏−𝑎

5
 Differential Calculus – One Variable

Generalized mean value theorem (Cauchy mean value theorem):

If 𝑓(𝑥) and 𝑔(𝑥) are two functions continuous in [a, b] and

differentiable in (a, b), and 𝑔’(𝑥) does not vanish anywhere inside
the interval then ∃ a point 𝑐 in (𝑎, 𝑏) such that
𝑓 ( 𝑏 ) − 𝑓 (𝑎 ) 𝑓 ′ (𝑐 )
=
𝑔(𝑏) − 𝑔(𝑎) 𝑔′(𝑐)
Proof: Define

𝑓 (𝑏 ) − 𝑓 (𝑎 )
𝜙(𝑥) = (𝑓(𝑥) − 𝑓 (𝑎)) − [ ] (𝑔(𝑥) − 𝑔(𝑎))
𝑔 (𝑏 ) − 𝑔 (𝑎 )

Note that 𝑔(𝑏) ≠ 𝑔(𝑎) because 𝑔′ does not vanish in (a, b). If
𝑔(𝑏) = 𝑔(𝑎) then Rolle’s Theorem implies 𝑔′ (𝑐 ) = 0, which
contradicts the assumption that 𝑔′ (𝑥) ≠ 0.

𝜙(𝑥) satisfies all hypotheses of the Rolle’s theorem on the interval

[a, b]. Then there exists a point 𝑐 ∈ (𝑎, 𝑏) such that 𝑐 ∈ (𝑎, 𝑏) and
𝜙 ′ (𝑐 ) = 0.

𝑓 (𝑏 ) − 𝑓 (𝑎 ) ′
⇒ 𝑓 ′ (𝑐 ) − [ ] 𝑔 (𝑐 ) = 0
𝑔 (𝑏 ) − 𝑔 (𝑎 )

𝑓 (𝑏 ) − 𝑓 (𝑎 ) 𝑓 ′ (𝑐 )
⇒[ ]= .
𝑔 (𝑏 ) − 𝑔 (𝑎 ) 𝑔′(𝑐)

Notice that:
𝑔(𝑥)=𝑥 𝑓(𝑥)=𝑓(𝑎)
Generalized MVT ⇒ Lagrange MVT ⇒ Rolle′ s Theorem

6
 Differential Calculus – One Variable

Ex: Using mean value theorem show that

|cos 𝑒 𝑥 − cos 𝑒 𝑦 | ≤ |𝑥 − 𝑦| for 𝑥, 𝑦 ≤ 0 (equality holds for x = y = 0)

Sol: Consider 𝑓 (𝑡) = cos 𝑒 𝑡 in the interval [𝑥, 𝑦]. Using Lagrange
mean value theorem
cos 𝑒 𝑥 − cos 𝑒 𝑦
= 𝑓 ′ (𝑐 ), 𝑐 ∈ (𝑥, 𝑦)
𝑥−𝑦
⇒ |cos 𝑒 𝑥 − cos 𝑒 𝑦 | ≤ |𝑥 − 𝑦| max 𝑓 ′ (𝑐 ) < |𝑥 − 𝑦|
𝑐∈(𝑥,𝑦)
as 𝑓 ′ (𝑡) = − et sin 𝑒 𝑡 ⇒ |𝑓 ′ (𝑡)| = |𝑒 | sin 𝑒 𝑡 | < 1 for 𝑡 < 0
𝑡 |

Ex: Using mean value theorem show that

𝑥
ln(1 + 𝑥) ≤ for 𝑥 > 0.
√(1+𝑥)

𝑡
Hint: Consider 𝑓(𝑡) = ln(1 + 𝑡) − in the interval [0, 𝑥].
√(1+𝑡)

𝑡
Also use the inequality 1 + ≥ √(1 + 𝑡)
2