§11.

3 The Integral Test and Estimates of Sums

In general, it is difficult to find the exact sum of a series (i.e. lim sn ) if it is not easy to find a simple
n!1
formula for the nth partial sum sn .
In the next few sections, we will develop various tests that can be used to determine whether a given
series converges or diverges without explicitly finding its sum.

X1
1 1 1 1 1
Example. Does the series 2
= 2 + 2 + 2 + 2 + · · · converge or diverge?
n=1
n 1 2 3 4

There is no simple formula for the sum sn of the

first n terms, but the computer-generated table of
approximate values (see right) suggests that the
partial sums are approaching ⇡ 1.64 as n ! 1.
That is, the series seems to be .

We can confirm this impression with a geometric argument:

X 1
1
The sum of the areas of the rectangles is exactly the sum of the series 2
.
n=1
n
If we exclude the first rectangle, the total area of the remaining rectangle is smaller
1
than the area under the curve y = 2 for x 1, which is equivalent to:
x

111
So all the partial sums must be less than

Hence we suspect that the sum of the areas of all the rectangles must be less than
X1
1
(i.e. 2
must be .)
n=1
n
Example. Determine the convergence of
X1
1
p (= )
n=1
n

The table of values (see right) suggests that the

partial sums are not approaching a finite number,
so this series seems to be .

Again, we confirm this with a geometric argument:

X1
1
The sum of the areas of all the rectangles is exactly the sum of the series p .
n=1
n

112
 1
The total area of these rectangles is greater than the area under the curve y = p
x
for x 1, which is equal to

That is, the area under the curve is . So the sum of the areas of all
the rectangles must be ; that is, this series is .

The same sort of geometric reasoning can be used to prove the following test:

The Integral Test

Let an = f (n) for all n k 1. Suppose that f (x) is a ,

and function on [k, 1). Then
1
X
and an
n=k

both converge or both diverge.

X1
1
Example. Does 2+1
converge or diverge?
n=1
n
1
Let f (x) = 2 , x 2 [1, 1).
x +1
First check that we can apply the Integral Test.
1. Is f decreasing on [1, 1)? Let’s check if f 0 < 0 on [1, 1):
✓ ◆
d 1
=
dx x2 + 1
2. Is f continuous on [1, 1)?

Recall that if a function is di↵erentiable on (a, b) then the function is con-

tinuous on (a, b). We found above that the derivative exists for all x 2 [1, 1).
That is, f is on [1, 1). So f is on [1, 1).
3. Is f positive on [1, 1)?

113
 Hence f is continuous, positive and decreasing on [1, 1), so we can apply the
Integral Test:

1
X 1
Since this integral is , we conclude that , by
n=1
n2 +1
the Integral Test.

1
X 1
Remark. The above result does NOT imply that the 2+1
converges to ⇡/4.
n=1
n
The Integral Test does not give us any information about what number the series
converges to.

114
p-Series
Definition. Let p > 0. A p-series is an infinite series of the form

X1
1
p
=
n=1
n

Example. The following are examples of p-series:

1
X 1 1 1
a) = 1 + + + · · · (p = ) - recall: this is the series
n=1
n 2 3
(we saw that this series is divergent)
X1
1 1 1
b) 2
= 1 + 2
+ 2
+· · · (p = ) (we saw that this series is convergent)
n=1
n 2 3

X1
1 1 1
c) p = 1+ p + p +· · · (p = ) (we saw that this series is divergent)
n=1
n 2 3

115
 X1
1
Example. For what values of p is the p-series p
convergent?
n=1
n
Case 1. p = 1 : This is the series, so it .
1
Case 2. p 6= 1 : Let f (x) = , x 2 [1, 1). Since f is continuous, positive
xp
and decreasing (check!), the Integral Test can be applied:
Z 1
1
dx =
1 xp

• If p > 1, then p + 1 , so
Z 1 ✓ p+1 ◆
1 t 1
dx = lim =
1 xp t!1 p+1 p+1
That is, when p > 1, the integral (and hence the series) by the
Integral Test.
• If 0 < p < 1, then p + 1 , so
Z 1 ✓ p+1 ◆
1 t 1
dx = lim =
1 xp t!1 p+1 p+1
That is, when 0 < p < 1, the integral (and hence the series) by
the Integral Test.
In summary,
X1
1
The p-series p
converges if and diverges otherwise.
n=1
n

Example. Does the following series converge or diverge?

1 1 1
1+ p
3
+p
3
+p
3
+ ···
2 3 4

This series is equivalent to .

This is a p-series with p = . Hence this series .

116
Estimating the Sum of a Series
P
Suppose that we used the Integral Test to show that the series an is convergent.
How can we get an approximation to the sum s of the series?

Using the same notation as the Integral Test, suppose that f (n) = an and f is
decreasing on [n, 1).

Recall:
1
X
• s= an = a1 + a2 + a3 + · · · + an + · · ·
n=1

• s n = a1 + a2 + a3 + · · · an
Definition. The remainder Rn denotes the error made when sn (the sum of the
first n terms) is used to approximate the sum s of a series. That is,

Rn =

There are two ways to express Rn using rectangles:

Z 1
Rn = an+1 + an+2 + · · · f (x) dx
n

Z 1
Rn = an+1 + an+2 + · · · f (x) dx
n+1

Based on the figures, we have the following error estimate.

117
 P f (n) = an , where f is a continuous, positive, decreasing function on [n, 1)
Suppose
and an converges to s. If Rn = s sn , then
Z 1 Z 1
f (x) dx < Rn < f (x) dx

Example.
X1
1
a) Estimate the error involved in approximating the sum of the series 3
by
n=1
n
using the sum of the first 10 terms, s10 .
Note: s10 = 113 + 213 + 313 + 413 + 513 + 613 + 713 + 813 + 913 + 1013 ⇡ 1.1975

1
One can show that f (x) = x3 is continuous, positive, and decreasing on [10, 1)
(check!). Now,

b) How many
p terms are required to ensure that the sum is accurate to within 0.0005?
Hint: 1000 ⇡ 31.6

Hence, we need terms to ensure accuracy to within 0.0005.

Homework. 11.3:# 3, 5, 7, 23, 25, 27, 29, 31, 37a, 39ad

118
§11.4 The Comparison Tests
Idea: Compare a given series with a series that is known to be convergent / divergent.

For each rectangle, an denotes the

Suppose that an , bn 0 for all n.
area of the dark portion and bn
P
denotes the combined area of the • if thePtotal area bn is finite, then the
white and dark portions. area an must also be .

P
• If the total area P an is infinite, then
the total area bn must also be
.

The (Direct) Comparison Test

1
X X1
Let an and bn be series and suppose that
n=1 n=1

0  an  bn for all n 1.
1
X 1
X
a) If the “bigger series” bn converges, the “smaller series” an .
n=1 n=1
1
X 1
X
b) If the “smaller series” an diverges, the “bigger series” bn .
n=1 n=1

In using the Comparison Test, we need some series whose convergence we know for
the purpose of comparison. We usually use one of these series:
P
• a p-series ( n1p converges if and diverges otherwise)
• a geometric series (converges if and diverges otherwise)
P
Remark. In testing many series an , we may find a suitable comparison series by
keeping only the (i.e. the terms that “grow” the fastest
as n ! 1) in the numerator and denominator of an .

119
Example. Determine whether the following series converges or diverges.
1
X 1
a)
n=1
2n2 + n + 1
For large n, the dominant term in the denominator is , so we com-
1
X
pare the given series with the series , which is a constant times a
n=1
series ( ), so it . Note that

1
X 1
We conclude that by the Comparison Test.
n=1
2n2 + n + 1
1
X 7
b) p 1
n=1
n 2

So the given series by the Comparison Test.

120
 1
X ln n
Example. Test the series for convergence.
n=1
n
3

1 y = ln x

1 2 3 4 5
1

3
1
X 1
is the series with the first terms
n=3
n
removed, so it is still divergent. Hence by the Comparison Test, the “larger” se-
X1 X1
ln n ln n
ries must . Since has two more terms added to the
n=3
n n=1
n
X1
ln n
divergent series , it must also .
n=3
n

121
The Limit Comparison Test

The following comparison is usually easier to apply than the Comparison Test.
1
X 1
X
Suppose that an and bn are series with positive terms. Let
n=1 n=1
⇢=

If <⇢< then the series both converge or both diverge.

Example. Test the series for convergence or divergence.

1
X 1
a)
n=1
2n 1
1
X
Take the comparison series of bn = , which is a
n=1
series ( ).
Remark. The Comparison Test will NOT work with this comparison series,
1 1
since n and the “smaller series” converges.
2 1 2n
an
⇢ = lim =
n!1 bn

1
X
Since 0 < ⇢ < 1, we can use the Limit Comparison Test. Since bn is
n=1
, we conclude that the original series is also by
the Limit Comparison Test.

122
 1
X n + 4n
b)
n=1
n + 6n
1
X
We use the comparison series of bn = , which is a
n=1
series ( ).

Now,
an
⇢ = lim = lim
n!1 bn n!1

Since 0 < ⇢ < 1, we can use the Limit Comparison Test.

X1
Since bn is , we conclude that the original series is also
n=1
by the Limit Comparison Test.
Remark. Comparison Test with the above comparison series does NOT work
n + 4n 4n
( > )
n + 6n 6n

Homework. 11.4: # 1, 7 - 33 (odd), 37

123
§11.5 Alternating Series
Up to now we only focused on series with non-negative terms. In this section, we discuss how to deal
with series whose terms are not necessarily all positive.

Definition. An alternating series is a series whose terms alternate between pos-

itive and negative numbers.
Example.
1
X 1 1 1
• ( 1)n+1 =1 + ···
n=1
n 2 3
1
X n 1 2 3
• ( 1)n = + + ···
n=1
n+1 2 3 4

Alternating Series Test (AST)

1
X
The alternating series an converges if:
n=1

(i) lim |an | = and

n!1

(ii) |a1 | |a2 | |a3 | ··· |an | · · · (i.e. |an | |an+1 | for all n 1)
1
X
Idea behind proof: Let the alternating series be ( 1)n+1 bn , where bn > 0
n=1

124
Example. Test the following alternating series for convergence.
1
X 1
a) ( 1)n+1 =
n=1
n
This is called an alternating harmonic series.
(i) lim |an | =
n!1

(ii) |an | =

✓ ◆0
1
Remark. Alternately, one can show f 0 (x) =  08x 2 [1, 1).
x

Hence the given series by the Alternating Series Test.

X1
( 1)n 3n
b)
n=1
4n 1
This is an alternating series. So we try the Alternating Series Test:
Condition (i): lim |an | =
n!1

That is, this series by the Test for Divergence.

Remark. In general, if condition (i) in the Alternating Series Test fails then the
series must diverge by the Test for Divergence.

125
 1
X
n+1 n2
c) ( 1)
n=1
n3 + 1

Check condition (i):

lim |an | =
n!1

Check condition (ii):

|an | |an+1 | FOR ALL n 1,

Hence the series by the Alternating Series Test.

126