Difference Equations Section 5.

3
to
Differential Equations Infinite Series Revisited

Recall from Section 1.3 that for a given sequence {an }, the sequence {sn } with nth term

sn = a1 + a2 + a3 + · · · + an (5.3.1)

is called an infinite series. An individual term sn is called a partial sum and we say the
series is convergent, or has a sum, if lim sn exists. If the series is not convergent, we say
n→∞
it is divergent. We write
∞
X
lim sn = a1 + a2 + a3 + · · · + an + · · · = an . (5.3.2)
n→∞
n=1

Example In Section 1.3 we saw that if an = rn , n = 0, 1, 2, . . ., then the associated

infinite series, called a geometric series, is convergent if and only if −1 < r < 1, in which
case
∞
X 1
rn = . (5.3.3)
n=0
1−r
For example,
∞
X 1 1 3
n
= 1 = .
n=0
3 1− 3
2

Geometric series comprise one of the few classes of series for which we can evaluate sums
exactly. For most series we can only approximate the sum by computing the partial sums
sn for sufficiently large values of n. However, before this procedure becomes meaningful, we
must first know that the series converges. Hence, in this section, as well as in Sections 5.4,
5.5, and 5.6, one of our primary goals will be the development of methods for determining
whether a given series converges or diverges.
We begin byP considering several
P∞ basic properties of infinite series. First, suppose we
∞
know that both n=1 an and n=1 bn are convergent series with
∞
X
an = L
n=1

and
∞
X
bn = M.
n=1

1 Copyright c by Dan Sloughter 2000

2 Infinite Series Revisited Section 5.3
P∞ P∞
If sn is the nth partial P
sum of n=1 an , tn is the nth partial sum of n=1 bn , and un is
∞
the nth partial sum of n=1 (an + bn ), then un = sn + tn . Thus

lim un = lim (sn + tn ) = lim sn + lim tn = L + M. (5.3.4)

n→∞ n→∞ n→∞ n→∞

That is,
∞
X ∞
X ∞
X
(an + bn ) = an + bn . (5.3.5)
n=1 n=1 n=1

P∞ P∞ P∞
Proposition If n=1 an and n=1 bn both converge, then n=1 (an + bn ) converges
and
∞
X ∞
X ∞
X
(an + bn ) = an + bn . (5.3.6)
n=1 n=1 n=1
P∞
Similarly, n=1 (an − bn ) converges and
∞
X ∞
X ∞
X
(an − bn ) = an − bn . (5.3.7)
n=1 n=1 n=1

Example From our results above, it follows that

∞   X ∞ ∞
X 1 1 1 X 1 1 1 3 5 11
n
+ n
= n
+ n
= 1 + 1 = + = .
n=0
3 5 n=0
3 n=0
5 1− 3 1− 5
2 4 4

P∞
Now suppose n=1 an is a convergent series,
∞
X
an = L,
n=1
P∞
and k is any constant.
P∞ If sn is the nth partial sum of n=1 an and tn is the nth partial
sum of the series n=1 kan , then tn = ksn . Hence

lim tn = lim ksn = k lim sn = kL. (5.3.8)

n→∞ n→∞ n→∞

That is,
∞
X ∞
X
kan = k an . (5.3.9)
n=1 n=1

P∞ P∞
Proposition If n=1 an converges and k is any constant, then n=1 an converges and
∞
X ∞
X
kan = k an . (5.3.10)
n=1 n=1
Section 5.3 Infinite Series Revisited 3

Example We have
∞ ∞ ∞  
X 10 10 X 1 X 1 1
= =5 =5 = 10.
n=1
2n 2 n=1 2n−1 n=0
2n 1 − 12

P∞ P∞
Notice that if n=1 an diverges, then n=1 kan must also diverge for any constant
P∞
k 6= 0. This follows because,
P∞ if, on the contrary, n=1 kan converged, then, by the previous
proposition, so would n=1 an since
∞ ∞
X X 1
an = (kan ). (5.3.11)
n=1 n=1
k

P∞ P∞
Proposition If n=1 an diverges, then n=1 kan diverges for any k 6= 0.
Example In Section 1.3 we saw that the harmonic series
∞
X 1
n=1
n

diverges. It follows that both

∞ ∞  
X 1 X 1 1
=
n=1
3n n=1 3 n

and
∞ ∞  
X 9 X 9 1
=
n=1
20n n=1
20 n

are divergent series.

It is also important to note that since
∞
X m−1
X ∞
X
an = an + an (5.3.12)
n=1 n=1 n=m

P∞ P∞
for any positive integer m, the series n=1 an converges if and only if the series n=m an
converges. In other words, convergence or divergence of a series is never determined by
the behavior of any finite number of terms.
Example It follows from the previous example that
∞
X 9
n=200
20n

diverges.
4 Infinite Series Revisited Section 5.3

Example The series

∞
X 3
n=4
5n

converges. Moreover,
∞ ∞   ∞  
X 3 X 3 1 3 X 1 3 1 3
n
= 4 n−4
= n
= 1 = .
n=4
5 n=4
5 5 625 n=0
5 625 1 − 5
500

P∞
Now suppose the series n=1 an converges with

∞
X
an = L.
n=1

P∞
Let sn = a1 + a2 + a3 + · · · + an be the nth partial sum of n=1 an . Now

an = sn − sn−1 ,
n
X ∞
X
lim sn = lim ai = ai = L,
n→∞ n→∞
i=1 i=1

and
n−1
X ∞
X
lim sn−1 = lim ai = ai = L,
n→∞ n→∞
i=1 i=1

Hence

lim an = lim (sn − sn−1 ) = lim sn − lim sn−1 = L − L = 0. (5.3.13)

n→∞ n→∞ n→∞ n→∞

That is, the nth term of a convergent series must have a limit of 0.
P∞
Proposition If n=1 an converges, then

lim an = 0. (5.3.14)
n→∞

Note that this result only demonstrates a consequence of a series converging, and
so does not provide a criterion to determine convergence. However, it may be useful in
showing that certain series are divergent. Namely, if either the sequence {an } does not
have a limit or
lim an 6= 0,
n→∞
P∞
then the series n=1 an must diverge. This result is often called the nth term test for
divergence.
Section 5.3 Infinite Series Revisited 5

Example The series

∞  
X 1
cos
n=1
n
diverges since  
1
lim cos = cos(0) = 1.
n→∞ n
P∞ n
Example The series n=1 (−1) diverges since {(−1)n } does not have a limit.
Example Note that
1
lim = 0,
n→ n
yet the harmonic series
∞
X 1
n=1
n
diverges.

p-series
In the next section we will consider a method for determining the convergence or divergence
of a series by comparing a given series with a series which is already known to converge or
diverge. In order to make significant use of such a result it is necessary to have a supply of
series whose convergence or divergence is already known. So far geometric series are the
only series we have studied in any detail. Now we will consider the class of series of the
form
∞
X 1
p
, (5.3.15)
n=1
n
where p is a fixed constant. Such series are called p-series. The following proposition
contains our main result.
Proposition The p-series
∞
X 1
, (5.3.16)
n=1
np
converges for p > 1 and diverges for p ≤ 1.
To demonstrate this result, we shall consider four cases. First, suppose p ≤ 0. Then

1 ∞, if p < 0,
lim p = (5.3.17)
n→∞ n 1, if p = 0.

Thus the series diverges by the nth term test for divergence.
Next, consider 0 < p < 1. Note that for any n > 0, the partial sum
1 1 1
sn = 1 + p
+ p + ··· + p, (5.3.18)
2 3 n
6 Infinite Series Revisited Section 5.3

0.8

0.6

0.4

0.2

2 4 6 8 10
Z 11
1
Figure 5.3.1 Rectangles for left-hand rule approximation for √ dx
1 x

is a left-hand rule approximation, using intervals of length 1, for the integral

Z n+1
1
dx. (5.3.19)
1 xp

1
See Figure 5.3.1 for the case p = 2 and n = 10. Since

1
f (x) =
xp

is a decreasing function on the interval [1, n + 1], sn is an upper sum for the integral
(5.3.19), and hence
Z n+1
1
sn ≥ dx. (5.3.20)
1 xp
Now
n+1 n+1
x1−p (n + 1)1−p − 1
Z
1
dx = = . (5.3.21)
1 xp 1−p 1 1−p
Thus
(n + 1)1−p − 1
sn ≥ . (5.3.22)
1−p
But, since 1 − p > 0,
(n + 1)1−p − 1
lim = ∞. (5.3.23)
n→∞ 1−p
Hence {sn } is an unbounded, increasing sequence, and so, from a result in Section 1.2,

lim sn = ∞. (5.3.24)
n→∞
Section 5.3 Infinite Series Revisited 7

0.8

0.6

0.4

0.2

2 4 6 8 10
Z 10
1
Figure 5.3.2 Rectangles for right-hand rule approximation for dx
1 x1.5

In other words,
∞
X 1
n=1
np

is a divergent series.
For p = 1, the p-series is the harmonic series and so diverges.
Finally, consider p > 1. If for any integer n > 1 we let

1 1 1 1
tn = p
+ p + p + ··· + p, (5.3.25)
2 3 4 n

then tn is a right-hand rule approximation, using intervals of length 1, for the integral

Z n
1
dx. (5.3.26)
1 xp

See Figure 5.3.2 for the case p = 1.5 and n = 10. Since

1
f (x) =
xp

is a decreasing function on the interval [1, n], tn is a lower sum for the integral (5.3.26),
and hence
Z n
1
tn ≤ p
dx. (5.3.27)
1 x
8 Infinite Series Revisited Section 5.3

Now Z n Z ∞
1 1
dx ≤ dx
1 xp 1 xp
Z n
1
= lim dx
n→∞ 1 xp
n
x1−p
= lim (5.3.28)
n→∞ 1 − p
1
n1−p − 1
= lim
n→∞ 1−p
1
= ,
p−1
where the final equality follows from the fact that, since p > 1,

1
lim n1−p = lim = 0. (5.3.29)
n→∞ n→∞ np−1

Thus
1
tn ≤ . (5.3.30)
p−1
Now if sn is the nth partial sum of
∞
X 1
p
,
n=1
n

then s1 = 1 and sn = 1 + tn , n = 2, 3, 4, . . .. Hence

1 p
sn ≤ 1 + = (5.3.31)
p−1 p−1

for n = 1, 2, 3, . . .. Thus {sn } is a bounded, increasing sequence, and so, from a result in
Section 1.2, must have a limit. That is, lim sn exists and
n→∞

∞
X 1
n=1
np

is a convergent series.
Example The series
∞
X 1
√
n=1
n

diverges because it is a p-series with p = 12 . Moreover, it now follows from our earlier
results that the series
∞
X 3
√
n=1
2 n
Section 5.3 Infinite Series Revisited 9

and
∞
X 1
√
n=10
n
both diverge as well.
Example The series
∞
X 1
n=1
n2
converges because it is a p-series with p = 2. Similar to our last example, it now follows
from our earlier results that the series
∞
X 35
n=1
6n2

and
∞
X 7
n=20
5n2
both converge as well. Moreover, from (5.3.31), we know that
∞
X 1 2
2
≤ = 2.
n=1
n 2 − 1

In Problem 5 in Section 4.6, you were asked to show that the integral
Z ∞
1
dx
1 xp
diverges for p < 1 and converges for p > 1. In Section 6.2 we will see that
Z ∞
1
dx
1 x
diverges as well (see also Problem 5 of this section). Combining these facts with our results
about p-series, it follows that
∞
X 1
n=1
np
converges if and only if Z ∞
1
dx
1 xp
converges. This should not be surprising considering the intimate relationship we have seen
between the partial sums of the series and the left-hand and right-hand rule approximations
for the integral. The essential ingredient in making these connections was that the function
1
f (x) =
xp
10 Infinite Series Revisited Section 5.3

is continuous, positive, and decreasing on the interval [1, ∞) when p > 0. In fact, it can be
shown, using arguments similar to those given above, that if g is a continuous, decreasing
function on [1, ∞) with g(x) > 0 for all x ≥ 1, then

∞
X
g(n)
n=1

converges if and only if

Z ∞
g(x)dx
1

converges. You are asked to verify this result, known as the integral test, in Problem 4.

Problems

1. For each of the following infinite series, decide whether the series converges or diverges
and explain your answer. If the series is a convergent geometric series, find its sum.
∞ ∞  
X 3 X 1
(a) (b) 4− n
n=0
5n n=0
2
∞   ∞
X 5 2 X
(c) n
+ n (d) (−3)n
n=1
2 3 n=1
∞   ∞  
X 1 X 1
(e) 1− (f) n sin
n=1
n n=1
n
∞ ∞  n
X (−1)n X 3
(g) (h) −
n=10
1000 n=2
7

2. For each of the following infinite series, decide whether the series converges or diverges
and explain your answer.
∞ ∞
X 4 X 2
(a) (b)
n=1
n3 n=1
n15
∞   ∞
X 1 5 X 3
(c) 2
− 4 (d) √
n=1
n n n=21
n
∞ ∞ r
X 1
X 1
(e) n− 3 (f)
n=1 n=1
n3
∞ ∞ √
X 1 X 210 − n
(g) √ (h)
n=3
234 n n=5
n2
Section 5.3 Infinite Series Revisited 11
P∞ P∞
3. Give an example of divergent series n=1 an and n=1 bn for which the series

∞
X
(an + bn )
n=1

converges.
4. Prove the integral test. That is, show that if g is a continuous decreasing function on
[1, ∞) with g(x) > 0 for all x ≥ 1, then
∞
X
g(n)
n=1

converges if and only if Z ∞

g(x)dx
1
converges.
5. Use the integral test to determine the convergence or divergence of each of the following.
∞ ∞
X 1 X 5
(a) 2
(b) √
n=1
n +3 n=2
n−1
Z ∞ ∞
1 X 3n
(c) dx (d) √
1 x n=2
n2 − 1
P∞
6. Find three different examples of divergent series n=1 an with the property that
lim an = 0.
n→∞

7. The following argument has been used to show that

∞
X 1
(−1)n = :
n=0
2

Let
∞
X
L= (−1)n .
n=0

Then
∞
X ∞
X ∞
X
n n
L= (−1) = 1 + (−1) = 1 − (−1)n = 1 − L.
n=0 n=1 n=0

Thus L = 1 − L, and so L = 12 . Where is the fallacy in this argument?