Exponential Function

Christian Jelo R. Artoza

May 12, 2020

1 Discussion
This discussion will be all about applying Taylor’s theorem for the exponential function ex and computing
problems involving exponentials. Pretty much the same content as the previous discussion but for the
exponential function.

1.1 Taylor’s Formula for the Exponential Function

Let f (x) = ex , a = 0, and b = x in Taylor’s formula. Then, as before we will first find the general expression
for the n-th derivative of f (x) taken at the point x = 0.

1.1.1 The n-th Derivative of f(x)

We know that f (n) (x) = ex for any nonnegative integer n, and e0 = 1. Thus, we have the general expression

f (n) (0) = 1 (for all n)

1.1.2 The Coefficient of the Taylor Polynomial

Let cn be the n-th coefficient of the Taylor polynomial for f . Then, it can be expressed as

1
cn =
n!

This is because of the fact that the n-th derivative of f taken at x = 0 is equal to 1 for all n.

1.1.3 Taylor’s Formula

By the assumptions stated before and the results from the preceding sections (the general expressions), we
have, by Theorem 1.1

x2 x3 xn
ex = 1 + x + + + ... + + Rn+1 (x)
2! 3! n!

Where
xn+1
Rn+1 = ec
(n + 1)!
where c is a point between 0 and x. Now that we have the formula, we will now talk about estimating that
remainder term, which will different from the sine and the cosine.

1
1.2 Estimate for the Remainder Term
the function f (x) = ex , the one we are currently interested in, is strictly increasing for all x. The statement
can be proven by showing that the derivative of f gives a positive value for all points x and can be observed
graphically.

This statement is important because the n-th derivative of the exponential function is
 equal to the func-
tion itself, ex , and so f (n) (x) is also strictly increasing for all x. Furthermore, f (n) (x) = f (n) (x) because
ex is always positive.

So there are two things to note when giving an estimate to the remainder term of  this Taylor formula:
(1) The fact that f (n) (x) is strictly increasing for all x and; (2) The fact that f (n) (x) = f (n) (x).

The first is useful in giving an idea to what is the maximum value of the remainder term. Since f (n) (x) is
increasing for all x, it’s maximum value over an interval will always be at one of the endpoints of the interval.
Hence, the maximum value of f (n) (x) over the interval between 0 and x is either taken at the point 0 or at
the point x. Let c be the point between 0 and x such that f (c) (c) is the maximum value. Then
(
(n) f (n) (x) if x > 0
f (c) =
1 if x < 0

x > 0 means we have the interval [0, x]. Thus, the maximum value of f (n) is taken at the point x. x < 0
means we have [x, 0]. Thus the maximum of f (n) is taken at the point 0, which is why it ended up being
equal to 1.

The upper bound Mn is usually (or initially) the maximum value of the n-th derivative between 0 and
x. So, we let Mn = f (n) (c), which is either taken at x or 0. Thus, by Theorem 2.1.,

n+1
|x|
|Rn+1 | ≤ ec
(n + 1)!

2
with
(
c ex if x > 0
e =
1 if x < 0.

1.3 EXAMPLE. Computing the number e

Let us compute the number e with an accuracy of 10−3 or up to 3 decimal places. This is simply done by
computing the value of the function ex at the point x = 1, and so we will use the Taylor formula we obtained
in section 1.1. Let’s deal with the remainder term first. Note the fact that e < 4. Now, since x = 1, we have

Mn = ec = e1 .

Thus,
1
|Rn+1 | ≤ e .
(n + 1)
And what we want is to have n such that
1
e ≤ 10−3 .
(n + 1)!

This is the point where we do trial and error. Let n = 4. Then,

1 1
e· ≤4· =≤ 10−3 .
5! 5!
Not even close. Let n = 6, then
1 1 4
4· =4· ≤ · 10−3 .
7! 5040 5
So by letting n = 6, our computed value will fall within the desired bounds of accuracy. We will now use
the Taylor formula in computing the value. We have

(1) (1) (1) (1) (1)

e = 1 + (1) + + + + + +E
2! 3! 4! 5! 6!
1 1 1 1 1
=1+1+ + + + + + E.
2 6 24 120 720
We can use a calculator to add up those terms. So, our final answer is

e = 2.718 + E

with
|E| ≤ 10−3

1.4 EXAMPLE. Knowing how many terms you need

How many terms of the Taylor formula do you need to compute e1/10 to an accuracy of 10−3 ? So, we are
not to find the value of e1/10 , but we just want to know what n is to compute it to an accuracy of 10−3 or
up to 3 decimals.

Since x = 1/10,
Mn = e1/10 ,

3
and so
1
|Rn+1 | ≤ e1/10 .
10n+1 (n + 1)!
And what we want is to have n such that
1
e1/10 ≤ 10−3 .
10n+1 (n + 1)!

But ex is strictly increasing, and so it is certain that e1/10 < 2 < e. Thus,
2
· 10−(n+1) ≤ 10−3 .
(n + 1)!

At this point we will do trial and error in finding the right n. Let n = 2. We have,
2
· 10−3 = 10−3 .
2!
Which fits within the desired bounds of accuracy. So, we need 2 Taylor polynomial terms plus the remainder
term, and so we need only

3 terms of the Taylor formula

to compute e1/10 to an accuracy of 10−3 .