Exponential and logarithms

MAT 137

Exponential and logarithms

Our textbook introduces exponential, logarithms, and their derivatives in a very dierent
manner which will not make sense just yet. These notes are intended to replace sections 7.27.5
from the textbook. (You may read those sections from the book if you wish, although you may
want to wait untill we study integration and the Fundamental Theorem of Calculus.). The notes
are a very short sketch and should be complemented with the lectures. The notes also contain
some facts about exponentials and logarithms that you already learned in high school.
To complement the notes, do the practice problems from sections 7.27.5 that are
posted on the course website.

Properties of exponentials

The expression ax makes sense whenever a > 0 and x 2 R . The following are common properties of
exponentials that you already know. For every a, b > 0 and every x, y 2 R :
a0 = 1
a1 = a
a

1
ax

ax ay = ax+y

ax
= ax
ay

(ax )y = axy
(ab)x = ax bx
When a > 1, the graph of y = ax looks like

Exponential and logarithms

MAT 137

6
5
4
3
2
1
-4

-3

-2

-1

-1
When 0 < a < 1, the graph of y = ax looks like

6
5
4
3
2
1
-4

-3

-2

-1
-1

Derivative of exponentials. Definition of e.

Let us fix a constant a > 0. Consider the function f (x) = ax . We want to compute the derivative of f from
the definition of derivative as a limit.
f (x + h)
h!0
h

f (x)

f 0 (x) = lim

Let us call La = lim

h!0

ah

1
h

ax+h ax
ax ah ax
ax (ah
= lim
= lim
h!0
h!0
h!0
h
h
h

= lim

1)

= ax lim

h!0

ah

1
h

. If this limit exists, then La simply represents a number. In that case, we have

d x
a = La ax
dx
In particular, f 0 (0) = La . We can try to compute this limit numerically. By plugging values of h close to 0,
it appears that
L2 = 0.6931 . . . ,
L3 = 1.0986 . . .
We also see that La , as a function of a, is increasing.
2

Exponential and logarithms

MAT 137

Definition. We define e as the only number e > 0 such that Le = 1. In other words, e is defined as the
d x
only number e > 0 such that
e = ex .
dx
For any other base a, we will determine the value of La in a little bit.

Definition of logarithm

Let a > 0, a 6= 1. Let x > 0. By definition, loga x is the only number y 2 R such that ay = x. In other
words:
loga x = y () ay = x
For example, if we want to compute log2 8, we are looking for a number y such that 2y = 8. The number
y = 3 works, so log2 8 = 3.
For now, this is going to be our definition of logarithm. There is a better definition, but it requires integration
(see section 7.2 of the textbook), so it will have to wait.
The definition means that logarithms are inverse functions of exponentials. This means that for every a > 0,
a 6= 1, x 2 R we have
loga ax = x,
and for every a > 0, a 6= 1, y > 0, we have

aloga y = y

The number e is one special basis for logarithms. For every x > 0, we write ln x for loge x.

Properties of logarithms

When a > 1, the graph of y = loga x looks like

4
3
2
1
-1

-1
-2
-3
-4
When 0 < a < 1, the graph of y = loga x looks like

Exponential and logarithms

MAT 137

4
3
2
1
-1

-1
-2
-3
-4
Here are some basic properties of logarithms. For every a > 0, a 6= 1, x, y > 0, b > 0, b 6= 1, r 2 R :
loga 1 = 0
loga a = 1
loga

1
=
x

loga x

loga (xy) = loga x + loga y

x
= loga x
y

loga y

loga xr = r loga x
loga x =

logb x
logb a

You should be able to prove every one of these properties from the definition of logarithm above, and from
the properties of exponentials. We will do one as an example. We suggest you try the rest as an exercise.
Theorem. Let a > 0, a 6= 1. Let x, y > 0. Then loga (xy) = loga x + loga y.
Proof. Let us call loga x = p and loga y = q. By definition of logarithm this means ap = x and aq = y. Now
we can compute the following:
ap+q = ap aq = xy
which, by definition of logarithm, means that loga (xy) = p + q = loga x + logz y.

Derivative of exponentials. Part 2

We can use logarithms to rewrite an arbitrary exponential as an exponential with base e. Specifically, for
every a > 0, x 2 R , we can write:
ax = ex ln a
(Why is this true?)
4

Exponential and logarithms

MAT 137

Lets go back to finding the derivative of an arbitrary exponential function. Let us fix a > 0 and let f (x) = ax .
Then
d x
d x ln a
d
f 0 (x) =
a =
e
= ex ln a (x ln a) = ex ln a ln a = ax ln a
dx
dx
dx
where we have used the chain rule.

The power rule

You already know the power rule. If c is a constant:

However, so far we were only able to prove it for certain values of c. We can now prove it for arbitrary values
of c (even for non-rational numbers!) We write xc as an exponential with base e, and then we use the chain
rule:
d c
d c ln x
d
1
c
x =
e
= ec ln x (c ln x) = ec ln x c = xc = cxc 1
dx
dx
dx
x
x

Derivatives of logarithms

Let us call g(x) = ln x. We want to find an expression for g 0 (x). We start by considering the identity:
x = eln x
We can dierentiate both sides with respect to x:
d
d ln x
x=
e
dx
dx
On the left-hand side, we get

d
dx x

= 1. On the right-hand side, using the chain rule, we get:

so that we get 1 = xf 0 (x) or f 0 (x) =

1
. In other words:
x
d
1
ln x =
dx
x

A similar calculation, left as an exercise to you, produces:

Logarithmic dierentiation

Example 1. Consider the function f (x) = (x2 + 1)x . We want to compute its derivative. There are at least
two ways to do this.

Exponential and logarithms

MAT 137

Method 1: Rewrite the function as an exponential with base e.

= ex ln(x

+1)

Second, we use the chain rule:

+1)

= ex ln(x

and the product rule:

+1)

ln(x2 + 1) + x

+1)

d
x ln(x2 + 1)
dx

2x
= x2 + 1
2
x +1

ln(x2 + 1) +

2x2
+1

x2

Method 2: Logarithmic dierentiation.

Let us call y = x2 + 1 . Taking logarithms on both sides, we get:

d
d
ln y =
x ln(x2 + 1)
dx
dx

and we use implicit dierentiation:

1 dy
2x
2x2
= ln(x2 + 1) + x 2
= ln(x2 + 1) + 2
y dx
x +1
x +1
from where we get

dy
2x
= y ln(x2 + 1) + x 2
dx
x +1

Example 2 Consider the function g(x) =

x2 + 3x
. We want to computer g 0 (x).
(1 + sin2 x)(x2 + 1)5

Method 1. We do not need any new method for computing this derivative. Just use the quotient rule,
the product rule, the power rule, and the chain rule repeatedly, and you will get the result. However,
this is long, tedious, and very error-prone. The second method is much more efficient.
Method 2. Call y = g(x). We can write
ln y =
so that

1
ln(x2 + 3x)
2

ln(1 + sin2 x)

d
d 1
ln y =
ln(x2 + 3x)
dx
dx 2

5 ln(x2 + 1)

ln(1 + sin2 x)

5 ln(x2 + 1)

and a much simpler calculation produces

2 sin x cos x
1 + sin2 x

dy
.
dx

2x
x2 + 1