eNote 3 1

eNote 3

Elementary Functions

In this eNote we will both repeat some of the basic properties for a selection of the (from high
school) well-known functions f ( x ) of one real variable x, and introduce some new functions,
which typically occur in a variety of applications. The basic questions concerning any function
are usually the following: How, and for which values of x, is the function defined? Which
values for f ( x ) do we get when we apply the functions to the x-elements in the domain? Is the
function continuous? What is the derivative f 0 ( x ) of the function – if it exists? As a new
concept, we will introduce a vast class of functions, the epsilon functions, which are denoted
by the common symbol ε( x ) and which we will use generally in order to describe continuity and
differentiability – also of functions of more variables, which we introduce in the following
eNotes.

(Updated: 22.9.2021 David Brander)

3.1 Domain and Range

In the description of a real function f ( x ) both the real numbers x where the function is
defined and the values that are obtained by applying the function on the domain are
stated. The Domain we denote D ( f ) and the range, or image, we denote R( f ).

Note: in higher mathematics, it is usual to define a function by specifying the domain

and codomain, (the set where the function in principle takes values) rather than the im-
age. For example: f : R → R given by f ( x ) = x2 . The codomain is R, but the range is
the set of non-negative numbers [0, ∞[⊂ R.
eNote 3 3.1 DOMAIN AND RANGE 2

Example 3.1 Some Domains and Ranges

Here are domains and the corresponding ranges for some well-known functions.

f1 (x) = exp( x ) , D( f1 ) = R = ] − ∞, ∞[ , R( f 1 ) = ]0, ∞[

f2 (x) = ln( x ) , D( f2 ) =]0, ∞[ , R( f 2 ) = R = ] − ∞, ∞[
√
f3 (x) = x , D( f3 ) = [0, ∞[ , R( f 3 ) = [0, ∞[
f4 (x) = x2 , D( f4 ) = R = ] − ∞, ∞[ , R( f 4 ) = [0, ∞[
(3-1)
f5 (x) = x7 + 8x3 + x − 1 , D( f5 ) = R = ] − ∞, ∞[ , R( f 5 ) = R = ] − ∞, ∞[
f6 (x) = exp(ln( x )) , D( f6 ) = ]0, ∞[ , R( f 6 ) = ]0, ∞[
f7 (x) = sin(1/x ) , D( f7 ) = ] − ∞, 0[∪]0, ∞[ , R( f 7 ) = [−1, 1]
f8 (x) = | x |/x , D( f8 ) = ] − ∞, 0[∪]0, ∞[ , R( f 8 ) = {−1} ∪ {1}

Figure 3.1: The well-known exponential function e x = exp( x ) and the natural loga-
rithmic function ln( x ). The red circles on the negative x-axis and at 0 indicate that the
logarithmic function is not defined on ] − ∞, 0].
eNote 3 3.1 DOMAIN AND RANGE 3

The function f 8 ( x ) in Example 3.1 is defined using | x |, which denotes the ab-
solute value of x, i.e.

 x>0, for x > 0
|x| = 0, for x = 0 (3-2)
− x > 0 , for x < 0 .


From this the domain and range for f 8 ( x ) follow directly.

Example 3.2 Tangent

The function
sin( x )
f ( x ) = tan( x ) = (3-3)
cos( x )
has the domain D ( f ) = R \ A, A denoting those real numbers x for which cos( x ) = 0, cos( x )
being the denominator, i.e.

D ( f ) = R \ { x | cos( x ) = 0} = R \ {(π/2) + p · π , p being an integer} . (3-4)

The range R( f ) is all real numbers, see Figure 3.2.

Figure 3.2: The graphs for the functions tan( x ) and cot( x ).
eNote 3 3.1 DOMAIN AND RANGE 4

Exercise 3.3

Let g( x ) denote the reciprocal function to the function tan( x ):

cos( x )
g( x ) = cot( x ) = (3-5)
sin( x )

Determine the domain for g( x ) and state it in the same way as above for tan( x ), see Figure
3.2.

3.1.1 Extension of the Domain to All of R

A function f ( x ) that is not defined for all real numbers can easily be extended to a func-
tion fb( x ), which has D ( fb) = R. One way of doing this is by the use of a curly bracket
in the following way:

Definition 3.4
Given a function f ( x ) with D ( f ) 6= R. We then define the 0-extension of f ( x ) by:

f ( x ) , for x ∈ D ( f )
f (x) = (3-6)
for x ∈ R \ D ( f ) .
b
0,

It is evident that depending on the application one can seal and extend the do-
main for f ( x ) in many other ways than choosing the constant 0 as the value for
the extended function at the points where the original function is not defined.

Naturally, the Range R( fb) for the 0-extended function is the original range for
f ( x ) united with 0, i.e. R( fb) = R( f ) ∪ {0} .

Hereafter we will assume – unless otherwise stated – that the functions we consider are
defined for all R possibly by extension as above.
eNote 3 3.2 EPSILON FUNCTIONS 5

3.2 Epsilon Functions

We introduce a special class of functions, which we will use in order to define the im-
portant concept of continuity.

Definition 3.5 Epsilon Functions

Every ε( x ) that is defined on an open interval that contains 0 and that assumes the
value ε(0) = 0 at x = 0 and moreover tends towards 0 when x tends towards 0
is called an epsilon function of x. Thus epsilon functions are characterized by the
properties:
ε(0) = 0 and ε( x ) → 0 for x → 0 . (3-7)
The last condition is equivalent to the fact that the absolute value of ε( x ) can be
made as small as possible by choosing the numerical value of x sufficiently small.
To be precise the condition means: For every number a > 0 there exists a number
b > 0 such that |ε( x )| < a for all x satisfying | x | < b.

The set of epsilon functions is very large:

Example 3.6 Epsilon Functions

Here are some simple examples of epsilon functions:

ε 1 (x) = x
ε 2 (x) = |x|
(3-8)
ε 3 ( x ) = ln(1 + x )
ε 4 ( x ) = sin( x ) .

The quality ’to be an epsilon function’ is rather stable: The product of an ep-
silon function and an arbitrary other function that only has to be bounded is
also an epsilon function. The sum and the product of two epsilon functions are
again epsilon functions. The absolute value of an epsilon function is an epsilon
function.
eNote 3 3.3 CONTINUOUS FUNCTIONS 6

Functions that are 0 in other places than x = 0 can also be epsilon functions:

If a function g( x ) has the properties g( x0 ) = 0 and g( x ) → 0 for x → x0 then

g( x ) is an epsilon function of x − x0 i.e. we can write g( x ) = ε g ( x − x0 ).

Exercise 3.7

Show that the 0-extension fb8 ( x ) of the function f 8 ( x ) = | x |/x is not an epsilon function. Hint:
If we choose k = 10 then clearly there does not exist a value of K such that

1 1
| f 8 ( x )| = | | x |/x | = 1 < , for all x with |x| < . (3-9)
10 K

Draw the graph for fb8 ( x ). This cannot be drawn without ’lifting the pencil from the paper’!

Exercise 3.8

Show that the 0-extension of the function f ( x ) = sin(1/x ) is not an epsilon function.

3.3 Continuous Functions

We can now formulate the concept of continuity by use of epsilon functions:

Definition 3.9 Continuity

A function f ( x ) is continuous at x0 if there exists an epsilon function ε f ( x − x0 ) such
that the following is valid on an open interval that contains x0 :

f ( x ) = f ( x0 ) + ε f ( x − x0 ) . (3-10)

If f ( x ) is continuous at every x0 on a given open interval in D ( f ) we say that f ( x )

is continuous on the interval.
eNote 3 3.4 DIFFERENTIABLE FUNCTIONS 7

Note that even though it is clear what the epsilon function precisely is in the
definition 3.9, viz. f ( x ) − f ( x0 ), then the only property in which we are in-
terested is the following: ε f ( x − x0 ) → 0 for x → x0 such that f ( x ) → f ( x0 )
for x → x0 , that is precisely as we know the concept of continuity from high
school!

Exercise 3.10

According to the above, all epsilon functions are continuous at x0 = 0 (with the value 0 at
x0 = 0). Construct an epsilon function that is not continuous at any of the points x0 = 1/n
where n = 1, 2, 3, 4, · · · .

Even though the concept of epsilon functions is central to the definition of

continuity (and as we shall see below, to the definition of differentiability),
epsilon functions need not be continuous for any other values than x0 = 0.

Exercise 3.11

Show that the 0-extension fb( x ) of the function f ( x ) = | x − 7|/( x − 7) is not continuous on
R.

3.4 Differentiable Functions

eNote 3 3.4 DIFFERENTIABLE FUNCTIONS 8

Definition 3.12 Differentiability

A function f ( x ) is differentiable at x0 ∈ D ( f ) if both a constant a and an epsilon
function ε f ( x − x0 ) exist such that

f ( x ) = f ( x0 ) + a · ( x − x0 ) + ( x − x0 ) · ε f ( x − x0 ) . (3-11)

It is the number a that we call f 0 ( x0 ) and it is well-defined in the sense that if f ( x )

can be stated at all in the form above (i.e. if f ( x ) is differentiable at x0 ) then there is
one and only one value for a that makes this formula true. With this definition of
the derivative f 0 ( x0 ) of f ( x ) at x0 we then have:

f ( x ) = f ( x0 ) + f 0 ( x0 ) · ( x − x0 ) + ( x − x0 ) · ε f ( x − x0 ) . (3-12)

If f ( x ) is differentiable for all x0 in a given open interval in D ( f ), we then naturally

say that f ( x ) is differentiable on the interval. We often write the derivative of f ( x )
at x in the following alternative way:

d
f 0 (x) = f (x) . (3-13)
dx

Explanation 3.13 The Derivative is Unique

We will show that there is only one value of a that fulfills Equation (3-11). Assume
that two different values, a1 and a2 both fulfill (3-11) possibly with two different
epsilon functions:

f ( x ) = f ( x0 ) + a1 · ( x − x0 ) + ( x − x0 ) · ε 1 ( x − x0 )
(3-14)
f ( x ) = f ( x0 ) + a2 · ( x − x0 ) + ( x − x0 ) · ε 2 ( x − x0 ) .
By subtracting (3-14) from the uppermost equation we get:

0 = 0 + ( a1 − a2 ) · ( x − x0 ) + ( x − x0 ) · (ε 1 ( x − x0 ) − ε 2 ( x − x0 )) , (3-15)

such that
a2 − a1 = ε 1 ( x − x0 ) − ε 2 ( x − x0 ) (3-16)
for all x 6= x0 – and clearly this cannot be true; the right hand side tends towards
0 when x tends towards x0 ! Therefore the above assumption, i.e. that a1 6= a2 , is
eNote 3 3.4 DIFFERENTIABLE FUNCTIONS 9

wrong. The two constants a1 and a2 must be equal, and this is what we should
realize.

The definition above is quite equivalent to the one we know from high school.
If we first subtract f ( x0 ) from both sides of the equality sign in Equation (3-12)
and then divide by ( x − x0 ) we get

f ( x ) − f ( x0 )
= f 0 ( x0 ) + ε f ( x − x0 ) → f 0 ( x0 ) for x → x0 , (3-17)
x − x0
i.e. the well-known limit value for the quotient between the increment in the
function f ( x ) − f ( x0 ) and the x-increment x − x0 . The reason why we do not
apply this known definition of f 0 ( x0 ) is simply that for functions of more vari-
ables the quotient does not make sense – but more about this in a later eNote.

Theorem 3.14 Differentiable Implies Continuous

If a function f ( x ) is differentiable at x0 , then f ( x ) is also continuous at x0 .

Proof

We have that
f ( x ) = f ( x0 ) + f 0 ( x0 ) · ( x − x0 ) + ( x − x0 ) ε f ( x − x0 )
(3-18)
= f ( x0 ) + f 0 ( x0 ) · ( x − x0 ) + ( x − x0 ) ε f ( x − x0 ) ,
 

and since the function in the square brackets on the right hand side is an epsilon function of
( x − x0 ) then f ( x ) is continuous at x0 .

But the opposite is not valid – here is an example that shows this:
eNote 3 3.4 DIFFERENTIABLE FUNCTIONS 10

Example 3.15 Continuous But Not Differentiable

The function f ( x ) = | x | is continuous but not differentiable at x0 = 0. The function is in itself

an epsilon function and therefore f ( x ) is continuous in 0. But now assume that there exist a
constant a and an epsilon function ε f ( x − x0 ) such that

f ( x ) = f ( x0 ) + a · ( x − x0 ) + ( x − x0 ) ε f ( x − x0 ). (3-19)

The following will then apply:

|x| = 0 + a · x + x · ε f (x) (3-20)

and hence for all x 6= 0:

|x|
= a + ε f (x) . (3-21)
x
If so a should both be equal to −1 and to 1 and this is impossible! Therefore the assumption
above that there exists a constant a is accordingly wrong; therefore f ( x ) is not differentiable.

Definition 3.16
The first degree approximating polynomial for f ( x ) expanded about the point x0 is
defined by:
P1,x0 ( x ) = f ( x0 ) + f 0 ( x0 ) · ( x − x0 ) . (3-22)

Note that P1,x0 ( x ) really is a first degree polynomial in x. The graph for the
function P1,x0 ( x ) is the tangent to the graph for f ( x ) at the point ( x0 , f ( x0 )),
see Figure 3.3. The equation for the tangent is y = P1,x0 ( x ), thus y =
f ( x0 ) + f 0 ( x0 ) · ( x − x0 ). The slope of the tangent is clearly α = f 0 ( x0 ) and
the tangent intersects the y-axis at the point (0, f ( x0 ) − x0 · f 0 ( x0 )). Later we
will find out how we can approximate with polynomials of higher degree n,
i.e. polynomials that are then denoted Pn,x0 ( x ).

3.4.1 Differentiation of a Product

eNote 3 3.4 DIFFERENTIABLE FUNCTIONS 11

Figure 3.3: Construction of the tangent y = P1,x0 ( x ) = f ( x0 ) + α · ( x − x0 ) with the

slope α = f 0 ( x0 ) for the function f ( x ). To the right the difference between f ( x ) and the
’tangent value’ P1,x0 ( x ).

Theorem 3.17 Differentiation of f ( x ) · g( x )

A product h( x ) = f ( x ) · g( x ) of two differentiable functions f ( x ) and g( x ) is differ-
entiable and its derivative is as follows:
d
( f ( x ) · g( x )) = f 0 ( x ) · g( x ) + f ( x ) · g0 ( x ) . (3-23)
dx

Even though this formula is rather well known from high school we shall give a short
sketch of a proof – to illustrate the use of epsilon functions.
eNote 3 3.4 DIFFERENTIABLE FUNCTIONS 12

Proof

Since f ( x ) and g( x ) are differentiable in x0 , we have:

f ( x ) = f ( x0 ) + f 0 ( x0 ) · ( x − x0 ) + ( x − x0 ) ε f ( x − x0 )
(3-24)
g ( x ) = g ( x0 ) + g 0 ( x0 ) · ( x − x0 ) + ( x − x0 ) ε g ( x − x0 ) ,

resulting in the product of the two right hand sides:

h( x ) = f ( x ) · g( x )
(3-25)
= f ( x0 ) · g( x0 ) + ( f 0 ( x0 ) · g( x0 ) + f ( x0 ) · g0 ( x0 )) · ( x − x0 ) + ( x − x0 )ε h ( x − x0 ) ,

where we have used ( x − x0 )ε h ( x − x0 ) as short for the remaining part of the product sum.
Furthermore any of the addends in the remaining part contains the factor ( x − x0 )2 or the
product of ( x − x0 ) with an epsilon function and therefore can be written in the stated form.
But then the product formula follows directly from the factor in front of ( x − x0 ) in Equation
(3-25):
h 0 ( x0 ) = f 0 ( x0 ) · g ( x0 ) + f ( x0 ) · g 0 ( x0 ) . (3-26)

3.4.2 Differentiation of a Quotient

The following differentiation rule is also well known from high school:

Theorem 3.18 Differentiation of f ( x )/g( x )

A quotient h( x ) = f ( x )/g( x ) involving two differentiable functions f ( x ) and g( x ),
is differentiable everywhere that g( x ) 6= 0, and the derivative is given in this well-
known fashion:
f 0 (x) f ( x ) · g0 ( x ) f 0 ( x ) · g( x ) − f ( x ) · g0 ( x )
 
d f (x)
= − = . (3-27)
dx g( x ) g( x ) g2 ( x ) g2 ( x )
eNote 3 3.4 DIFFERENTIABLE FUNCTIONS 13

Exercise 3.19

Use the epsilon function argument in the same way as in the differentiation rule for a product
to show Equation 3.18.

3.4.3 Differentiation of Composite Functions

Theorem 3.20 The Chain Rule for Composite Functions

A function h( x ) = f ( g( x )) that is composed of two differentiable functions f ( x ) and
g( x ) is in itself differentiable at every x0 with the derivative

h0 ( x0 ) = f 0 ( g( x0 )) · g0 ( x0 ) (3-28)

Proof

We exploit that the two functions f ( x ) and g( x ) are differentiable. In particular g( x ) is dif-
ferentiable at x0 :
g( x ) = g( x0 ) + g0 ( x0 )( x − x0 ) + ( x − x0 ) · ε g ( x − x0 ) , (3-29)
and the function f (u) is differentiable at u0 = g( x0 ):
f (u) = f (u0 ) + f 0 (u0 )(u − u0 ) + (u − u0 ) · ε f (u − u0 ) . (3-30)
From this we get, setting u = g( x ) and u0 = g( x0 ):
h( x ) = f ( g( x ))
= f ( g( x0 )) + f 0 ( g( x0 ))( g( x ) − g( x0 ) + ( g( x ) − g( x0 ) · ε f ( g( x ) − g( x0 )
= h( x0 ) + f 0 ( g( x0 ))( g0 ( x0 )( x − x0 ) + ( x − x0 ) · ε g ( x − x0 )) (3-31)
0
+ ( g ( x0 )( x − x0 ) + ( x − x0 ) · ε g ( x − x0 )) · ε f ( g( x ) − g( x0 )
= h( x0 ) + f 0 ( g( x0 )) g0 ( x0 ) · ( x − x0 ) + ( x − x0 ) · ε h ( x − x0 ) ,
from which we directly read that h0 ( x0 ) = f 0 ( g( x0 )) g0 ( x0 ) – because this is exactly the unique
coefficient of ( x − x0 ) in the above expression.


eNote 3 3.5 INVERSE FUNCTIONS 14

Exercise 3.21

Above we have used – at the end of Equation (3-31) – that

f 0 ( g( x0 )) · ε g ( x − x0 ) + ( g0 ( x0 ) + ·ε g ( x − x0 )) · ε f ( g( x ) − g( x0 )) (3-32)

is an epsilon function, which we accordingly can call (and have called) ε h ( x − x0 ). Consider
why this is entirely OK.

Exercise 3.22

Find the derivatives of the following functions for every x-value in their respective domains:

f 1 ( x ) = ( x2 + 1) · sin( x )
f 2 ( x ) = sin( x )/( x2 + 1) (3-33)
2
f 3 ( x ) = sin( x + 1) .

3.5 Inverse Functions

The exponential function exp( x ) and the logarithmic function ln( x ) are inverse func-
tions to each other – as is well known the following is valid:

exp(ln( x )) = x for x ∈ D (ln) = ]0, ∞[ = R(exp)

(3-34)
ln(exp( x )) = x for x ∈ D (exp) = ] − ∞, ∞[ = R(ln) .

Note that even though exp( x ) is defined for all x, the inverse function ln( x ) is
only defined for x > 0 – and vice versa (!).

The function f ( x ) = x2 has an inverse function in its respective intervals of monotony,

i.e. where f ( x ) is either increasing or decreasing: The inverse function on the interval
√
[0, ∞[ where f ( x ) is increasing is the well-known function g( x ) = x. Thus the function
f ( x ) maps the interval A = [0, ∞[ one-to-one onto the interval B = [0, ∞[, and the
eNote 3 3.5 INVERSE FUNCTIONS 15

inverse function g( x ) maps the interval B one-to-one onto the interval A such that:
√
f ( g( x )) = ( x )2 = x for x ∈ B = [0, ∞[
√ (3-35)
g( f ( x )) = x2 = x for x ∈ A = [0, ∞[ .

The inverse function to f ( x ) on the interval ] − ∞, 0] where f ( x ) is decreasing is the

√
function h( x ) = − x, which is not defined on the same interval as f ( x ). The function
f ( x ) maps the interval C =] − ∞, 0] one-to-one onto the interval D = [0, ∞[, and the
inverse function h( x ) maps the interval D one-to-one onto the interval C such that:
√
f (h( x )) = (− x )2 = x for x ∈ D = [0, ∞[
√ (3-36)
h( f ( x )) = − x2 = x for x ∈ C = ] − ∞, 0] .

If f ( x ) is not monotonic on an interval, it means that we can obtain the same

function-value f ( x ) for more x-values – in the same way as x2 = 1 both for x = 1
and for x = −1, and then the function is not one-to-one on the interval. The
functions cos( x ) and sin( x ) are only monotonic on certain subintervals of the
x-axis, see Figure 3.7. If we wish to define inverse functions to the functions
we must choose the interval with care, see Section 3.8 and Figure 3.8.

Definition 3.23 Notation for Inverse Functions

We denote the inverse function for a given function f ( x ) by f ◦−1 ( x ). The inverse
function is generally defined by the following properties on suitably chosen inter-
vals A and B that are part of D ( f ) and D ( f ◦−1 ), respectively

f ◦−1 ( f ( x )) = x for x ∈ A ⊂ D( f )
(3-37)
f ( f ◦−1 ( x )) = x for x ∈ B ⊂ D ( f ◦−1 ) .

We use here the symbol f ◦−1 ( x ) in order to avoid confusion with ( f ( x ))−1 =
1/ f ( x ). However the reader should note that the standard notation is simply
f −1 for the inverse function. The graph for the inverse function g( x ) = f ◦−1 ( x )
to a function f ( x ) can be obtained by mirroring the graph for f ( x ) in the diag-
onal in the first quadrant in the ( x, y)-coordinate system – i.e. the line with the
equation y = x – see Figure 3.4.
eNote 3 3.5 INVERSE FUNCTIONS 16

Figure 3.4: The graph for a function f ( x ) and the graph for the inverse function g( x ).
It is valid that g( x ) = f ◦−1 ( x ) and f ( x ) = g◦−1 ( x ), but they each have their own
definition intervals.

3.5.1 Differentiation of Inverse Functions

Theorem 3.24 Differentiation of Inverse Functions

If a differentiable function f ( x ) has the inverse function f ◦−1 ( x ) and if
f 0 ( f ◦−1 ( x0 )) 6= 0, then the inverse function f ◦−1 ( x ) is itself differentiable at x0 :

1
( f ◦−1 )0 ( x0 ) = (3-38)
f 0 ( f ◦−1 ( x 0 ))
eNote 3 3.6 HYPERBOLIC FUNCTIONS 17

Proof

From the definition of inverse functions we have

h( x ) = f ( f ◦−1 ( x )) = x , (3-39)

so h0 ( x0 ) = 1, but we also have from the chain rule in (3-28):

h0 ( x0 ) = f 0 ( f ◦−1 ( x0 )) · ( f ◦−1 )0 ( x0 ) = 1 , (3-40)

from which we get the result by dividing by f 0 ( f ◦−1 ( x0 )).

3.6 Hyperbolic Functions

Definition 3.25 Hyperbolic Cosine and Hyperbolic Sine

We will define two new functions cosh( x ) and sinh( x ) as the unique solution to the
following system of differential equations with initial conditions. The two solutions
are denoted hyperbolic cosine and hyperbolic sine, respectively:

cosh0 ( x ) = sinh( x ) , cosh(0) = 1

(3-41)
sinh0 ( x ) = cosh( x ) , sinh(0) = 0 .

The names cosh( x ) and sinh( x ) (often spoken as “cosh” and “sinsh”) look like cos( x )
and sin( x ), but the functions are very different, as we shall demonstrate below.

Yet there are also fundamental structural similarities between the two pairs of functions
and this is what motivates the names. In the system of differential equations for cos( x )
eNote 3 3.6 HYPERBOLIC FUNCTIONS 18

and sin( x ) only a single minus sign separates this from (3-41):

cos0 ( x ) = − sin( x ) , cos(0) = 1

0 (3-42)
sin ( x ) = cos( x ) , sin(0) = 0 .

In addition (again with the decisive minus sign as the only difference) the following
simple analogy to the well-known and often used relation cos2 ( x ) + sin2 ( x ) = 1 applies:

Theorem 3.26 Fundamental Relation of cosh( x ) and sinh( x )

cosh2 ( x ) − sinh2 ( x ) = 1 . (3-43)

Proof

Make the derivative with respect to x on both sides of the equation (3-43) and conclude that
cosh2 ( x ) − sinh2 ( x ) is a constant. Finally use the initial conditions.

Exercise 3.27

Show directly from the system of differential equations (3-41) that the two ”new” functions
are in fact not so new:
e x + e− x
cosh( x ) = , D (cosh) = R , R(cosh) = [1, ∞[
2 (3-44)
e x − e− x
sinh( x ) = , D (sinh) = R , R(sinh) = ] − ∞, ∞[
2
eNote 3 3.6 HYPERBOLIC FUNCTIONS 19

Figure 3.5: Hyperbolic cosine, cosh( x ), and hyperbolic sine, sinh( x ).

Exercise 3.28

Show directly from the expressions found in Exercise 3.27, that

cosh2 ( x ) − sinh2 ( x ) = 1 . (3-45)

Exercise 3.29

The graph for the function f ( x ) = cosh( x ) looks a lot like a parabola, viz. the graph for the
function g( x ) = 1 + ( x2 /2) when we plot both functions on a suitably small interval around
x0 = 0. Try this! If we instead plot the two graphs in very large x-interval, we learn that
the two functions have very different graphical behaviours. Try this, i.e. try to plot both
functions on the interval [−50, 50]. Comment upon and explain the qualitative differences.
Similarly compare the two functions sinh( x ) and x + ( x3 /6) in the same way.

It is natural and useful to define hyperbolic analogies to tan( x ) and cot( x ). This is done
as follows:
eNote 3 3.6 HYPERBOLIC FUNCTIONS 20

Definition 3.30 Hyperbolic Tangent and Hyperbolic Cotangent

sinh( x ) e2x − 1
tanh( x ) = = 2x , D (tanh) = R , R(tanh) = ] − 1, 1[
cosh( x ) e +1

(3-46)
cosh( x ) e2x + 1
coth( x ) = = 2x , D (coth) = R − {0} ,
sinh( x ) e −1
R(coth) = ] − ∞, −1[ ∪ ]1, ∞[ .

Figure 3.6: Hyperbolic tangent, tanh( x ), and hyperbolic cotangent, coth( x ).

The derivatives of cosh( x ) and of sinh( x ) are already given by the defining system in
(3-41).

d
cosh( x ) = sinh( x )
dx
d
sinh( x ) = cosh( x )
dx
d 1 (3-47)
tanh( x ) = = 1 − tanh2 ( x )
dx cosh2 ( x )
d −1
coth( x ) = 2
= 1 − coth2 ( x ) .
dx sinh ( x )
eNote 3 3.7 THE AREA FUNCTIONS 21

Exercise 3.31

Show the last two expressions for the derivatives for tanh( x ) and coth( x ) in (3-47) by the use
of the differentiation rule in Theorem 3.18.

3.7 The Area Functions

The inverse functions to the hyperbolic functions are called area functions and are
named cosh◦−1 ( x ) = arcosh( x ), sinh◦−1 ( x ) = arsinh( x ), tanh◦−1 ( x ) = artanh( x ), and
coth◦−1 ( x ) = arcoth( x ), respectively.

Since the functions cosh( x ), sinh( x ), tanh( x ), and coth( x ) all can be expressed in terms
of exponential functions it is no surprise that the inverse functions and their derivatives
can be expressed by logarithmic functions. We gather the information here:

p
arcosh( x ) = ln( x + x2 − 1) for x ∈ [1, ∞[
p
arsinh( x ) = ln( x + x2 + 1) for x ∈ R
 
1 1+x (3-48)
artanh( x ) = ln for x ∈ ] − 1, 1[
2 1−x
x−1
 
1
arcoth( x ) = ln for x ∈ ] − ∞, 1[ ∪ ]1, ∞[ .
2 x+1
d 1
arcosh( x ) =√ for x ∈]1, ∞[
dx x2 −1
d 1
arsinh( x ) =√ for x∈R
dx x2 +1 (3-49)
d 1
artanh( x ) = for x ∈ ] − 1, 1[
dx 1 − x2
d 1
arcoth( x ) = for x ∈ ] − ∞ 1[ ∪ ]1, ∞[ .
dx 1 − x2

3.8 The Arc Functions

The inverse functions to the trigonometric functions are a bit more complicated. As
mentioned earlier here we must choose for each trigonometric function an interval
eNote 3 3.8 THE ARC FUNCTIONS 22

Figure 3.7: Cosine and Sine Functions.

where the function in question is monotonic. In return, once we have chosen such an
interval, it is clear how the inverse function should be defined and how it should then
be differentiated. The inverse functions to cos( x ), sin( x ), tan( x ), and cot( x ) are usu-
ally written arccos( x ), arcsin( x ), arctan( x ), and arccot( x ), respectively; their names are
arccosine, arcsine, arctangent, and arccotangent. As above we gather the results here:

cos◦−1 ( x ) = arccos( x ) ∈ [0, π ] for x ∈ [−1, 1]

sin◦−1 ( x ) = arcsin( x ) ∈ [−π/2, π/2] for x ∈ [−1, 1]
(3-50)
tan◦−1 ( x ) = arctan( x ) ∈ [−π/2, π/2] for x∈R
cot◦−1 ( x ) = arccot( x ) ∈]0, π [ for x∈R .
d −1
arccos( x ) =√ for x ∈] − 1, 1[
dx 1 − x2
d 1
arcsin( x ) =√ for x ∈] − 1, 1[
dx 1 − x2 (3-51)
d 1
arctan( x ) = for x ∈R
dx 1 + x2
d −1
arccot( x ) = for x ∈R .
dx 1 + x2

Note that the derivatives for arccos( x ) and arcsin( x ) are not defined at x0 = 1
or at x0 = −1. This is partly because, if the function we consider is only defined
on a bounded interval then we cannot say that the function is differentiable
at the end-points of the interval. Moreover the formulas for arccos0 ( x ) and
arcsin0 ( x ) show that they are not defined at x0 = 1 or x0 = −1; these values
give 0 in the denominators.
eNote 3 3.8 THE ARC FUNCTIONS 23

Figure 3.8: The arccosine function is defined here.

Exercise 3.32

Use a suitable modification of arctan( x ) in order to determine a new differentiable (and hence
continuous) function f ( x ) that looks like the 0-extension of | x |/x (which is neither continuous
nor differentiable), i.e. we want a function f ( x ) with the following properties: 1 > f ( x ) >
0.999 for x > 0.001 while −0.999 > f ( x ) > −1 for x < −0.001. See Figure 3.10. Hint: Try to
plot arctan(1000x ).
eNote 3 3.8 THE ARC FUNCTIONS 24

Figure 3.9: Arccosine and arcsine. Again the red circles indicate that the arc-functions
are not defined outside the interval [−1, 1]. Similarly the green circular disks indicate
that the arc-functions are defined at the end-points x = 1 and x = −1.

Figure 3.10: The arctangent function.

eNote 3 3.9 SUMMARY 25

3.9 Summary

We have treated some of the fundamental properties of some well-known and some
not so well-known functions. How are they defined, what are their domains, are they
continuous, are they differentiable, and if so what are their derivatives?

• A function f ( x ) is continuous at x0 if f ( x ) − f ( x0 ) is an epsilon function of ( x −

x0 ), i.e.
f ( x ) = f ( x0 ) + ε f ( x − x0 ) . (3-52)

• A function f ( x ) is differentiable at x0 with the derivative f 0 ( x0 ) if

f ( x ) = f ( x0 ) + f 0 ( x0 )( x − x0 ) + ( x − x0 )ε f ( x − x0 ) .

• If a function is differentiable at x0 , then it is also continuous at x0 . The converse

does not apply.

• The derivative of a product of two functions is

d
( f ( x ) · g( x )) = f 0 ( x ) · g( x ) + f ( x ) · g0 ( x ) . (3-53)
dx

• The derivative of a quotient of two functions is

f 0 (x) f ( x ) · g0 ( x ) f 0 ( x ) · g( x ) − f ( x ) · g0 ( x )
 
d f (x)
= − = . (3-54)
dx g( x ) g( x ) g2 ( x ) g2 ( x )

• The derivative of a composite function is

d
f ( g( x )) = f 0 ( g( x )) · g0 ( x ) . (3-55)
dx

• The derivative of the inverse function f ◦−1 ( x ) is

 0 1
f ◦−1 (x) = . (3-56)
f 0 ( f ◦−1 ( x ))