Chapter 3

Derivatives

3.1 Derivative of a function. Differentiation rules

3.1.1 Slope of a curve
The slope of a curve at a point P is a measure of the steepness of the curve. If Q is a point
on the curve near P , then the slope of the curve at P is approximately the slope of the line
segment P Q. The slope of the curve at P is defined to be the limit of the slope of P Q as
Q approaches P along the curve.

In symbols, slope of curve at P = limQ→P (slope of P Q).

To find the slope of the curve y = x2 at point P = (1, 1) we choose a point Q on the
curve near P . Let the x–coordinate of Q be 1 + h with h small. The y–coordinate of Q is
(1 + h)2 . We now calculate

(1 + h)2 − 1
slope of P Q = = 2 + h.
(1 + h) − 1

As Q approaches P , h approaches 0. Thus:

slope of curve at (1, 1) = lim (2 + h) = 2.

h→0

1
Definition 3.1.1. The derivative of function f at point c, f ′ (c), is the slope of the curve
y = f (x) at point (c, f (c)), that is:
f (c + h) − f (c)
f ′ (c) = lim ,
h→0 h
whenever the limit exists.
We shall say that f is differentiable at point c if f ′ (c) exists.

f (c + h) − f (c)

P
h

O c c+h

3.1.2 Table of derivatives of basic elementary functions

1. (xα )′ = αxα−1 (α is any number).
2. (ln x)′ = x1 .
3. (ax )′ = ax ln a, in particular (ex )′ = ex .
4. (sin x)′ = cos x.
5. (cos x)′ = − sin x.
1
6. (tan x)′ = , (x ̸= π2 + πn, n integer).
cos2 x
1
7. (arcsin x)′ = √ (−1 < x < 1).
1 − x2
1
8. (arccos x)′ = − √ (−1 < x < 1).
1 − x2
1
9. (arctan x)′ = .
1 + x2

3.1.3 The line tangent to a curve

The line tangent to a curve at a point is defined to be the line that passes through the point
and that has slope that is the same as the slope of the curve at that point. Thus,
y − f (c) = f ′ (c)(x − c)
is the the equation of the line tangent to y = f (x) at point P = (c, f (c)).

2
Example 3.1.2.
√
1. Find the line tangent to y = x at (16, 4).

1 1
Solution: f (x) = x1/2 , f ′ (x) = x−1/2 , f ′ (16) = .
2 8
1 1
Hence, y − 4 = (x − 16), or y = x + 2.
8 8
2. Find the line tangent to y = |x| at (0, 0).

Solution: There is no tangent line to y = |x| at (0, 0), since the function f (x) = |x|
is not differentiable at 0. To see this, notice that the limit

f (0 + h) − f (0) |h|
lim = lim
h→0 h h→0 h

does not exist (the left limit is −1 and the right limit is 1).

3.1.4 One–sided derivatives

If there is the limit
f (c + h) − f (c) f (c + h) − f (c)
 
lim lim ,
h→0 + h h→0 − h

then it is called the right–hand (left–hand ) derivative of the function f at the point c and
is denoted f ′ (c+ ) (f ′ (c− )).

Theorem 3.1.3. f ′ (c) exists if and only if both f ′ (c+ ) and f ′ (c− ) exists and they are equal.
In this case, f ′ (c) = f ′ (c+ ) = f ′ (c− ).
 2
x , if x ≤ 0;
Example 3.1.4. Is the function f (x) = differentiable at 0?
xe−1/x , if x > 0.

Solution: Yes, and f ′ (0) = 0.

f (0 + h) − f (0) h2
f ′ (0− ) = lim = lim = 0;
h→0− h h→0− h
f (0 + h) − f (0) he−1/h
f ′ (0+ ) = lim = lim = e−∞ = 0.
h→0+ h h→0+ h

3.1.5 Continuity and differentiability

Continuity is a necessary condition for differentiability. In other words, a discontinuous
function cannot be differentiable.

Theorem 3.1.5. Let f be differentiable at c. Then, it is continuous at c.

3
Proof. By assumption the limit

f (c + h) − f (c)
f ′ (c) = lim
h→0 h
exists. We want to prove that f is continuous

at c, that is, limh→0 f (c + h) = f (c) or, equiv-
alently, that limh→0 f (c + h) − f (c) = 0. To this end consider the following computations:

h f (c + h) − f (c)
= 0 · f ′ (c) = 0.


lim f (c + h) − f (c) = lim h · lim
h→0 h h→0 h→0 h

ax − x2 , if x < 1;

Example 3.1.6. Discuss the differentiability of the function f (x) = ,
b(x − 1), if x ≥ 1.
where a, b ∈ R.

Solution: First we study continuity. The domain of f is the whole real line. For x < 1
and x > 1 it is given by elementary functions, which are continuous. It remains to consider
the frontier point x = 1. We have f (1) = 0 and limx→1− f (x) = limx→1− ax − x2 = a − 1.
Thus, f is continuous at 1 if and only if a = 1. Hence

If x ̸= 1, then f is continuous for any a, b and at x = 1, f is continuous if and

only if a = 1 (b arbitrary).

Now we go with differentiability. Clearly, f is differentiable at any point x ̸= 1. When

a ̸= 1 f is not differentiable at x = 1 since it is not continuous at this point. Hence, let us
consider a = 1.
f (1 + h) − f (1) (1 + h) − (1 + h)2 − 0
f ′ (1− ) = lim = lim
h→0− h h→0− h
2
(1 + h) − (1 + 2h + h ) −h − h2
= lim = lim = −1;
h→0− h h→0− h
f (1 + h) − f (1) b(1 + h − 1)
f ′ (1+ ) = lim = lim = b.
h→0 + h h→0+ h

Hence f ′ (1− ) = f ′ (1+ ) = f ′ (1) if and only if b = −1. In summary

If x ̸= 1, then f is differentiable for any a, b and at x = 1, f is differentiable if

and only if a = 1 and b = −1.

3.1.6 Rules of differentiation

Let f and g be functions differentiable at point c. Then, the sum, difference, product by
a scalar, product and quotient are also differentiable at c and the derivatives are given by
the following expressions.

1. Sum: (f + g)′ = f ′ + g ′ ;

2. Difference: (f − g)′ = f ′ − g ′ ;

4
 3. Product by a scalar: (λf )′ = λf ′ , λ ∈ R;

4. Product: (f · g)′ = f ′ · g + f · g ′ ;
 ′
f f ′g − f g′
5. Quotient: = , g(c) ̸= 0.
g g2

3.1.7 Chain Rule

(Derivative of a compose function). Let f be differentiable at c and let g be differentiable
at f (c). Then the composition g ◦ f is differentiable at c and the derivative

(g ◦ f )′ (c) = g ′ (f (c)) · f ′ (c).

Example 3.1.7. Find the derivative of y = sin (3x + x3 ).

Solution: We can represent the function in the form y = sin t where t = 3x +x3 . Using
the chain rule we get

y ′ = (sin t)′ |t=3x +x3 (3x + x3 )′ = cos (3x + x3 )(3x ln 3 + 3x2 ).

√
Example 3.1.8. Find the derivative of h(x) = ex − x2 at the point c = 1.

Solution:
 1 ′
 1 1 1 1
h′ (x) = (ex − x2 ) 2 = (ex − x2 )− 2 (ex − x2 )′ = (ex − x2 )− 2 (ex − 2x).
2 2
Hence h′ (1) = e−2
√
2 e−1
, approximately 0.274
Example 3.1.9. In the following examples it is supposed that f is a differentiable function.

′
ˆ ef (x) = f ′ (x)ef (x) ;

′
ˆ af (x) = (ln a)f ′ (x)af (x) .

f ′ (x)
ˆ (ln f (x))′ = ;
f (x)
f ′ (x)
ˆ (arctan f (x))′ = .
1 + f 2 (x)

3.1.8 Derivative of the inverse function

Let f be continuous and one-to-one in an open interval (x − δ, x + δ) and such that f ′ (x)
exists. Then f −1 is differentiable at y = f (x) and the derivative is

′ 1 1
f −1 (y) = = . (3.1.1)
f ′ (x) f ′ (f −1 (y))

5
 The proof of this assertion is easy using the Chain Rule. For, deriving in both sides of
the identity
x = f −1 (f (x)),
we obtain
′
1 = f −1 (f (x)) · f ′ (x).
The formula follows, once we substitute y = f (x).
1
Example 3.1.10. Prove that arctan′ x = .
1 + x2
Solution: The function arctan x is the inverse of the function tan x. According to the
formula (1.1.1) above
1
arctan′ y = ,
1 + tan2 x
because tan′ x = 1 + tan2 x, and where y = tan x. Thus,
1
arctan′ y = .
1 + y2
Of course, we can change the name of the variable from y to x to get the result.

3.1.9 Using the derivative for approximations

The line tangent to a curve at a point (c, f (c)) coincides with the curve at the point of
tangency, and constitute a good approximation of the curve at points near (c, f (c)). In
fact, a function is differentiable at a point when the graph of the function at this point can
be well approximated by a straight line (the tangent line).
Thus, for small values of h, the value of f (c + h) can be approximated by the known
valued of f (c) and f ′ (c):
f (c + h) ≈ f (c) + f ′ (c)h. (3.1.2)
Another way to express this is by increments : the increase in the value of f, △y, is
approximately the increment in x, h = △x, multiplied by f ′ (c):

△y ≈ f ′ (c)△x
√
Example 3.1.11. Without using a pocket calculator, give an approximated value of 0.98.

√
Solution: Let us consider the function f (x) = 1 + x. Notice that f (0) = 1,
√ 1
f (−0.02) = 0.98, f ′ (x) = (1 + x)−1/2 , f ′ (0) = 0.5. We find from formula (1.1.2)
2
with c = 0 and h = −0.02 that
√
0.98 = f (0 − 0.02) ≈ f (0) + f ′ (0)(−0.02) = 1 + 0.5(−0.02) = 0.99.
√
Example 3.1.12. Without using a pocket calculator, give an approximated value of 177.

6
 √
Solution: Let us consider the function f (x) = x. As base point c, one that would
be a perfect square (in order to compute its square root immediately) and that is as close
to 177 as it is possible (in
√ order that the increment would be as least as possible). The
candidate is c = 169, as 169 = 13. The derivative of f at c = 169 is
1 1
f ′ (169) = √ =
2 169 26
. √ √
So 177 = 169 + 8 = f (169 + 8) ≈ f (169) + f ′ (169).8 = 13 + 8
26 ≈ 13.307.
The correct value with 3 decimals is 13.304

3.1.10 Implicit differentiation

Definition 3.1.13. An equation F (x, y) = 0 defines y = f (x) in an implicit way near the
point (x0 , y0 ) when it is satisfied that:
1. F (x0 , y0 ) = 0

2. if (x, y) is close to the point (x0 , y0 ) : F (x, y) = 0 ⇐⇒ y = f (x).

Theorem 3.1.14. F (x0 , y0 ) = 0, ∂F
∂y (x0 , y0 ) ̸= 0 =⇒
The equation F (x, y) = 0 defines y = f (x) in an implicit way near the point (x0 , y0 ) .
Theorem 3.1.15. y0′ = f ′ (x0 ) can be obtained from the equation:
∂F ∂F ′
∂x (x0 , y0 ) + ∂y (x0 , y0 )y0 = 0 (∗)
This way, even if we do not know the explicit expression of y = f (x), we can have an
approximate idea of the function knowing that
y − y0 = f ′ (x0 )(x − x0 )
is the tangent line of f (x) at the point (x0 , y0 ).
Besides, if by taking the derivative of the equation (∗) we find y0 ” ̸= 0, our information
on the function improves since:

1. if y0 ” > 0 =⇒ f is convex near x0 =⇒ the graph of f lies above the tangent line near
the point (x0 , y0 ).

2. if y0 ” < 0 =⇒ f is convex near x0 =⇒ the graph of f lies below the tangent line near
the point (x0 , y0 ).

3.2 Some Theorems on Differentiable Functions

3.2.1 Monotonicity
The function f is said to be increasing at a point c if there is some interval around the point
c in which f (x) > f (c) for x > c, f (x) < f (c) for x < c.
A decrease of a function at a point can be defined analogously.
For example, c = 0 is a point of increasing of x3 , but is not of x2 .
Theorem 3.2.1. If the function f is differentiable at c and f ′ (c) > 0 (f ′ (c) < 0), then f
increases (decreases) at the point c.

7
 The theorem establishes only a sufficient condition, because c = 0 is a point of increasing
of f (x) = x3 but f ′ (0) = 0.

The following results are about the monotonic behavior of a differentiable function in
an interval.

Theorem 3.2.2. For the function f differentiable on an interval I to be increasing (de-

creasing) it is necessary and sufficient that for every x ∈ I f ′ (x) ≥ 0 (f ′ (x) ≤ 0) holds.

Theorem 3.2.3. If f ′ (x) > 0 (f ′ (x) < 0) for every x ∈ I, then f is strictly increasing
(decreasing) on the interval I.

Example 3.2.4. Find the intervals of monotonicity of f (x) = 3x − x3 .

Solution: We have f ′ (x) = 3 − 3x2 = 3(1 − x2 ). Since f ′ (x) > 0 for x ∈ (−1, 1) and
f ′ (x) < 0 for x ∈ (−∞, 1) and x ∈ (1, +∞), it follows that f is strictly increasing in [−1, 1]
and strictly decreasing in (−∞, −1] ∪ [1, ∞).

3.2.2 Local Extremum of functions

Derivatives are a very useful tool for locating and identifying maximum and minimum values
(extremum) of functions. In the following, we suppose that the function f is defined in an
open interval (c − δ, c + δ) around c.

Definition 3.2.5. The function f has a local maximum (minimum) at the point c if there
is δ > 0 such that for every x ∈ (c − δ, c + δ)

f (x) ≤ f (c) (f (x) ≥ f (c)).

A local maximum or a local minimum are local extremum of f .

Theorem 3.2.6. If the function f has an extremum at the point c, then the derivative f ′ (c)
is either zero or does not exist.

Proof. Without loss of generality, suppose that c a local minimum point of f and that f ′ (c)
exists. By definition of a local minimum, we have f (c + h) ≥ f (c) for every h with |h| < δ.
Let h > 0 and consider the quotient

f (c + h) − f (c)
.
h
It is non–negative and the limit exists when h → 0 and equals f ′ (c), since f is differentiable
at c. Given that the limit of non–negative quantities must be non–negative, we get the
inequality f ′ (c) ≥ 0. Consider now h < 0. Then, the above quotient is non–positive.
Taking the limit as h → 0 we get the reverse inequality f ′ (c) ≤ 0. Hence it must be
f ′ (c) = 0 and we are done.

The points where the function is not differentiable or the derivative vanishes are possible
extrema of f , and for this reason they are called critical points of f .

8
Example 3.2.7. Find the critical points of f (x) = 3x − x3 and g(x) = |x|.

Solution: Function f is differentiable at every point and f ′ (x) = 3(1 − x2 ). Thus,

f ′ (x)= 0 if and only if x = ±1. Hence, the critical points of f are 1 and −1. Function g is
differentiable at every point except c = 0, where it has a corner. Actually, the derivative

′ 1, if x > 0;
g (x) =
−1, if x < 0,
never vanishes. Hence, 0 is the only critical point of g.
Theorem 3.2.8. Suppose that f is differentiable in an interval I = (c − δ, c + δ) around
point c (except, maybe, for the point c itself ). Then, if the derivative of f changes sign from
plus to minus (from minus to plus) when passing through the point c, then f has a local
maximum (minimum) at the point c. If the derivative does not change sign when passing
through the point c, then the function f does not posses an extremum at the point c.
Example 3.2.9. Find the the local extrema points of f (x) = 3x − x3 and g(x) = |x|.

Solution: We know from Example 1.2.4 that the sign of f ′ change from minus to plus
at −1 and from plus to minus at 1, hence f has a local minimum at −1, and f has a local
maximum at 1. On the other hand, g ′ change from minus to plus at 0 (see Example 1.2.7),
hence although g is not differentiable at 0, g has a local minimum at this point.

3.2.3 Theorems of Rolle and Lagrange

Theorem 3.2.10 (Rolle’s Theorem). Let the function f satisfy the following conditions:
1. f is continuous on [a, b];
2. f is differentiable in (a, b);
3. f (a) = f (b).
Then there is a point c ∈ (a, b) such that f ′ (c) = 0.
Rolle’s Theorem states that there is a point c ∈ (a, b) such that the tangent line to the
graph of the function f at the point (c, f (c)) is parallel to the x–axis.
Theorem 3.2.11 (Lagrange’s Theorem). Let the function f satisfy the following conditions:
1. f is continuous on [a, b];
2. f is differentiable in (a, b).
Then there is a point c ∈ (a, b) such that
f (b) − f (a) = f ′ (c)(b − a).
Lagrange’s Theorem is also known as Mean Value Theorem. It can be interpreted as
follows: The number
f (b) − f (a)
b−a
is the slope of the line r which passes through the points (a, f (a)) and (b, f (b)) of the graph
of f , and f ′ (c) is the slope of the tangent to the graph of f at (c, f (c)). Lagrange’s formula
shows that this tangent line is parallel to the straight line r.

9
3.3 L’Hopital’s Rule
Now we present a useful technique for evaluating limits that uses the derivatives of the
functions involved.
Theorem 3.3.1 (Indetermination of the type 0/0). Assume that the following conditions
are fulfilled:
1. The functions f and g are defined and differentiable in an interval I = (c − δ, c + δ)
around point c (except, maybe, the point c itself );

2. limx→c f (x) = limx→c g(x) = 0;

3. The derivative g ′ (x) ̸= 0 for any x ∈ I (except, maybe, the point c itself ).
f ′ (x)
4. There exists the limit limx→c .
g ′ (x)
Then,
f (x) f ′ (x)
lim = lim ′ .
x→c g(x) x→c g (x)

sin ax
Example 3.3.2. Evaluate limx→0 , where a, b ∈ R.
tan bx
Solution: The limit is of the indeterminate type 0/0. It is easy to verify that all
conditions of Theorem 1.3.1 are fulfilled. Consequently,
sin ax a cos ax a
lim = lim b
= .
x→0 tan bx x→0 b
cos2 bx

Theorem 3.3.3 (Indetermination of the type ±∞/∞). Assume that (1), (3) and (4) of
Theorem 1.3.1 are fulfilled and that (2) is replaced by
(2’) limx→c f (x) = limx→c g(x) = ±∞.
Then,
f (x) f ′ (x)
lim = lim ′ .
x→c g(x) x→c g (x)

Example 3.3.4. Evaluate limx→0+ x ln x.

ln x
Solution: The limit is of the indeterminate type 0 · ∞. Writing x ln x as we get
1/x
an indeterminate form ∞/∞. Applying L’Hospital Rule one obtains
ln x 1/x
lim = lim = lim (−x) = 0.
x→0+ 1/x x→0+ −1/x2 x→0+
Remark 3.3.5. Similarly to Theorems 1.3.1 and 1.3.3, L’Hopital’s Rule can be also stated
when x → +∞ or x → −∞,
f (x) f ′ (x)
lim = lim ′ .
x→±∞ g(x) x→±∞ g (x)

10
 ln x
Example 3.3.6. Evaluate limx→∞ .
x
Solution: The limit is of the indeterminate type ∞/∞. Differentiating above and
below, we get
ln x 1/x
lim = lim = 0.
x→∞ x x→∞ 1

Remark 3.3.7. Indeterminate forms of other types, 0 · ∞, ∞ − ∞, 1∞ , 00 or ∞0 can be

reduced to indeterminate forms of type 0/0 or ∞/∞ and to them we can apply L’Hopital’s
Rule.
Example 3.3.8. Evaluate limx→∞ x1/x .

Solution: The limit is of the indeterminate type ∞0 . We represent x1/x = eln x/x and
ln x 1
study limx→∞ = limx→∞ = 0, hence the limit is e0 = 1.
x x
 
x 1
Example 3.3.9. Evaluate limx→1 − .
x − 1 ln x

Solution: The limit is of indeterminate type ∞ − ∞. If we combine fractions, then

we obtain
x ln x − (x − 1)
,
(x − 1) ln x
which is of the form 0/0 at x = 1. Differentiating above and below, we get

ln x
1 − x−1 + ln x
which is again 0/0 at x = 1. Another differentiation above and below gives

x−1 x
−2 −1
=
x +x 1+x
which has limit 1/2 as x → 1. Thus, we had to apply L’Hospital Rule twice to find that
the limit is 1/2.
When the hypotheses of the theorems are not satisfied, we might obtain incorrect an-
swers, as in the following example.
ln x −∞
Example 3.3.10. Clearly, limx→0+ = + = −∞. If we attempt to use L’Hopital
x 0
Rule, we would obtain
ln x x−1
lim = lim = +∞,
x→0+ x x→0+ 1

which is incorrect. Notice that the limit is not indeterminate, hence theorems 1.3.1 and
1.3.3 do not apply.

11
3.4 Optimization of continuous functions on intervals [a, b]
Consider a continuous function f defined on an interval I = [a, b]. By Weierstrass’ Theorem,
f attains in [a, b] global extremum. On the other hand, since a global extremum is also a
local extremum, if the global extremum are in the open interval (a, b), then they must be
critical points of f . Hence, to locate and classify the global extremum of f , we will use the
following recipe:

1. Find the critical points of f in (a, b);

2. Evaluate f at the critical points found in (a) and at the extreme points of the interval,
a, b;

3. Select the maximum value (global maximum) and the minimum value (global mini-
mum).

Example 3.4.1. Find and classify the extremum points of f (x) = 3x − x3 in the interval
[−2, 2].

Solution: Since f is continuous and I = [−2, 2] is closed and bounded, by Weierstrass’

Theorem f attains on I global maximum and minimum. Then, as explained above, the
possible global extremum are among the critical points of f in I and the extreme points
of the interval I: −2 and 2. We know that −1 ∈ I is a local minimizer, f (−1) = −2, and
1 ∈ I is a local maximizer, f (1) = 2, see Example 1.2.9. On the other hand, f (−2) = 2 and
f (2) = −2, thus points −1 and 2 are both global minimizers of f in I, and points −2 and
1 are global maximizers of f in I.

12