December 30, 2008 11:14 m91-ch09 Sheet number 1 Page number 596 cyan magenta yellow black

9
INFINITE SERIES
9.1 SEQUENCES
DEFINITION OF A SEQUENCE
Stated informally, an infinite sequence, or more simply a sequence, is an unending suc-
cession of numbers, called terms. It is understood that the terms have a definite order; that
is, there is a first term a1 , a second term a2 , a third term a3 , a fourth term a4 , and so forth.
Such a sequence would typically be written as
a1 , a2 , a3 , a4 , . . .
where the dots are used to indicate that the sequence continues indefinitely. Some specific
examples are
1, 2, 3, 4, . . . , 1, 12 , 13 , 14 , . . . ,
2, 4, 6, 8, . . . , 1, −1, 1, −1, . . .
Each of these sequences has a definite pattern that makes it easy to generate additional
terms if we assume that those terms follow the same pattern as the displayed terms. However,
such patterns can be deceiving, so it is better to have a rule or formula for generating the
terms. One way of doing this is to look for a function that relates each term in the sequence
to its term number. For example, in the sequence
2, 4, 6, 8, . . .
each term is twice the term number; that is, the nth term in the sequence is given by the
formula 2n. We denote this by writing the sequence as
2, 4, 6, 8, . . . , 2n, . . .
We call the function f(n) = 2n the general term of this sequence. Now, if we want to know
a specific term in the sequence, we need only substitute its term number in the formula for
the general term. For example, the 37th term in the sequence is 2 · 37 = 74.

Example 1 In each part, find the general term of the sequence.

1 2 3 4 1 1 1 1
(a) , , , ,...
2 3 4 5
(b) , , , ,...
2 4 8 16

(c) 1
2
, − 23 , 34 , − 45 , . . . (d) 1, 3, 5, 7, . . .
December 30, 2008 11:14 m91-ch09 Sheet number 2 Page number 597 cyan magenta yellow black

Solution (a). In Table 9.1.1, the four known terms have been placed below their term
term numbers, from which we see that the numerator is the same as the term number and the
... ...
number 1 2 3 4 n denominator is one greater than the term number. This suggests that the nth term has
numerator n and denominator n + 1, as indicated in the table. Thus, the sequence can be
term 1 2 3 4 ... n ...
2 3 4 5 n+1
expressed as
1 2 3 4 n
, , , ,..., ,...
2 3 4 5 n+1
Table 9.1.2 Solution (b). In Table 9.1.2, the denominators of the four known terms have been ex-
term pressed as powers of 2 and the first four terms have been placed below their term numbers,
1 2 3 4 ... n ...
number from which we see that the exponent in the denominator is the same as the term number.
This suggests that the denominator of the nth term is 2n , as indicated in the table. Thus, the
term 1 1 1 1 ... 1 ...
2 22 23 24 2n sequence can be expressed as
1 1 1 1 1
, , , ,..., n,...
2 4 8 16 2

Solution (c). This sequence is identical to that in part (a), except for the alternating signs.
Thus, the nth term in the sequence can be obtained by multiplying the nth term in part (a)
by (−1)n+1 . This factor produces the correct alternating signs, since its successive values,
starting with n = 1, are 1, −1, 1, −1, . . . . Thus, the sequence can be written as
Table 9.1.1
1 2 3 4 n
, − , , − , . . . , (−1)n+1 ,...
2 3 4 5 n+1

Table 9.1.3 Solution (d). In Table 9.1.3, the four known terms have been placed below their term
numbers, from which we see that each term is one less than twice its term number. This
term
number 1 2 3 4
... n ... suggests that the nth term in the sequence is 2n − 1, as indicated in the table. Thus, the
sequence can be expressed as
term 1 3 5 7 . . . 2n − 1 . . .
1, 3, 5, 7, . . . , 2n − 1, . . .

When the general term of a sequence

a1 , a2 , a3 , . . . , an , . . . (1)

is known, there is no need to write out the initial terms, and it is common to write only the
general term enclosed in braces. Thus, (1) might be written as
{an }+ ⬁
n=1 or as {an }⬁n=1
For example, here are the four sequences in Example 1 expressed in brace notation.
A sequence cannot be uniquely deter-
mined from a few initial terms. For
example, the sequence whose general
sequence brace notation
term is

f(n) = 13 (3 − 5n + 6n2 − n3 ) 1 , 2, 3, 4, . . . , n , . . . n +∞
2 3 4 5 n+1 n + 1 n=1
has 1, 3, and 5 as its first three terms, 1 , 1, 1, 1 , . . . , 1 , . . . 1 +∞
but its fourth term is also 5. 2 4 8 16 2n 2n n=1
1 , − 2 , 3 , − 4 , . . . , (−1) n+1 n , . . . n +∞
(−1) n+1
2 3 4 5 n+1 n+1 n=1
∞
1, 3, 5, 7, . . . , 2n − 1, . . . {2n − 1}+n=1
December 30, 2008 11:14 m91-ch09 Sheet number 4 Page number 599 cyan magenta yellow black

9.1.1 definition A sequence is a function whose domain is a set of integers.

Typically, the domain of a sequence is the set of positive integers or the set of nonnegative
integers. We will regard the expression {an }+ ⬁
n=1 to be an alternative notation for the function
f(n) = an , n = 1, 2, 3, . . . , and we will regard {an }+ ⬁
n=0 to be an alternative notation for the
function f(n) = an , n = 0, 1, 2, 3, . . . .

LIMIT OF A SEQUENCE
Since sequences are functions, we can inquire about their limits. However, because a
sequence {an } is only defined for integer values of n, the only limit that makes sense is the
limit of an as n → +⬁. In Figure 9.1.2 we have shown the graphs of four sequences, each
of which behaves differently as n → +⬁:

• The terms in the sequence {n + 1} increase without bound.

• The terms in the sequence {(−1)n+1 } oscillate between −1 and 1.
• The terms in the sequence {n/(n + 1)} increase toward a “limiting value” of 1.
  n 
• The terms in the sequence 1 + − 12 also tend toward a “limiting value” of 1, but
do so in an oscillatory fashion.

y y y y
7
6 1
5
4 n 1 1
3 1 2 3 4 5 6 7 8 9
1
2 2
1 −1
n n n
1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9

+∞ +∞ n +∞ +∞
n+1 (−1)n+1 1 +
− 12 n
n=1 n=1 n+1 n=1 n=1

Figure 9.1.2

9.1.2 definition A sequence {an } is said to converge to the limit L if given any
How would you define these limits?
 > 0, there is a positive integer N such that |an − L| <  for n ≥ N. In this case we
lim an = +⬁ write
n → +⬁
lim an = L
lim an = −⬁ n → +⬁
n → +⬁
A sequence that does not converge to some finite limit is said to diverge.

Example 2 The first two sequences in Figure 9.1.2 diverge, and the second two con-
verge to 1; that is,
n   n 
lim = 1 and lim 1 + − 12 =1
n → +⬁ n + 1 n → +⬁

The following theorem, which we state without proof, shows that the familiar properties
of limits apply to sequences. This theorem ensures that the algebraic techniques used to
find limits of the form lim can also be used for limits of the form lim .
x → +⬁ n → +⬁
December 30, 2008 11:14 m91-ch09 Sheet number 5 Page number 600 cyan magenta yellow black

9.1.3 theorem Suppose that the sequences {an } and {bn } converge to limits L1 and
L2 , respectively, and c is a constant. Then:
(a) lim c = c
n → +⬁

(b) lim can = c lim an = cL1

n → +⬁ n → +⬁

(c) lim (an + bn ) = lim an + lim bn = L1 + L2

n → +⬁ n → +⬁ n → +⬁

(d ) lim (an − bn ) = lim an − lim bn = L1 − L2

n → +⬁ n → +⬁ n → +⬁
Additional limit properties follow from
(e) lim (an bn ) = lim an · lim bn = L1 L2
those in Theorem 9.1.3. For example, n → +⬁ n → +⬁ n → +⬁
use part (e) to show that if an → L and
m is a positive integer, then an lim an L1
n → +⬁
m m
(f) lim = = (if L2 = 0)
lim (an ) = L
n → +⬁
n → +⬁ bn lim bn L2
n → +⬁

If the general term of a sequence is f(n), where f(x) is a function defined on the entire
interval [1, +⬁), then the values of f(n) can be viewed as “sample values” of f(x) taken

y at the positive integers. Thus,

L
if f(x) → L as x → +⬁, then f(n) → L as n → +⬁
f (3)
f (2)
(Figure 9.1.4a). However, the converse is not true; that is, one cannot infer that f(x) → L
as x → +⬁ from the fact that f(n) → L as n → +⬁ (Figure 9.1.4b).
f (1)
x
1 2 3 4 5 6 7 8
Example 3 In each part, determine whether the sequence converges or diverges by
If f (x) → L as x → + ∞, examining the limit as n → +⬁.
then f (n) → L as n → + ∞.
 +⬁
+⬁
n n+1 n
(a) (a) (b) (−1)
2n + 1 n=1 2n + 1 n=1

+⬁
y 1
(c) (−1)n+1 (d) {8 − 2n}+ ⬁
n=1
L n n=1
f (x)
Solution (a). Dividing numerator and denominator by n and using Theorem 9.1.3 yields
n 1 lim 1 lim 1
n → +⬁ n → +⬁
lim = lim = =
x n → +⬁ 2n + 1 n → +⬁ 2 + 1/n lim (2 + 1/n) lim 2 + lim 1/n
1 2 3 4 5 6 7 8 n → +⬁ n → +⬁ n → +⬁
1 1
f (n) → L as n → + ∞, but f (x) = =
2+0 2
diverges by oscillation as x → + ∞.
Thus, the sequence converges to 12 .
(b)
Figure 9.1.4
Solution (b). This sequence is the same as that in part (a), except for the factor of
(−1)n+1 , which oscillates between +1 and −1. Thus, the terms in this sequence oscillate
between positive and negative values, with the odd-numbered terms being identical to those
in part (a) and the even-numbered terms being the negatives of those in part (a). Since the
sequence in part (a) has a limit of 12 , it follows that the odd-numbered terms in this sequence
approach 12 , and the even-numbered terms approach − 12 . Therefore, this sequence has no
limit—it diverges.
December 30, 2008 11:14 m91-ch09 Sheet number 6 Page number 601 cyan magenta yellow black

Solution (c). Since 1/n → 0, the product (−1)n+1 (1/n) oscillates between positive and
negative values, with the odd-numbered terms approaching 0 through positive values and
the even-numbered terms approaching 0 through negative values. Thus,
1
lim (−1)n+1 =0
n → +⬁ n
so the sequence converges to 0.

Solution (d). lim (8 − 2n) = −⬁, so the sequence {8 − 2n}+ ⬁

n=1 diverges.
n → +⬁

Example 4 In each part, determine whether the sequence converges, and if so, find
its limit.
1 1 1 1
(a) 1, , 2 , 3 , . . . , n , . . . (b) 1, 2, 22 , 23 , . . . , 2n , . . .
2 2 2 2

Solution. Replacing n by x in the first sequence produces the power function (1/2)x , and
replacing n by x in the second sequence produces the power function 2x . Now recall that if
0 < b < 1, then bx → 0 as x → +⬁, and if b > 1, then bx → +⬁ as x → +⬁ (Figure 0.5.1).
Thus, 1
lim = 0 and lim 2n = +⬁
2n
n → +⬁ n → +⬁

So, the sequence {1/2n } converges to 0, but the sequence {2n } diverges.

n +⬁
Example 5 Find the limit of the sequence .
en n=1

Solution. The expression n

lim
en
n → +⬁

is an indeterminate form of type ⬁/⬁, so L’Hôpital’s rule is indicated. However, we cannot

apply this rule directly to n/en because the functions n and en have been defined here only at
the positive integers, and hence are not differentiable functions. To circumvent this problem
we extend the domains of these functions to all real numbers, here implied by replacing n
by x, and apply L’Hôpital’s rule to the limit of the quotient x /ex . This yields
x 1
lim x = lim x = 0
x → +⬁ e x → +⬁ e

from which we can conclude that

n
lim =0
n → +⬁ en

√
n
Example 6 Show that lim n = 1.
n → +⬁

Solution. √
lim n
n = lim n1/n = lim e(1/n) ln n = e0 = 1 By L’Hôpital’s rule
applied to (1/x) ln x
n → +⬁ n → +⬁ n → +⬁

Sometimes the even-numbered and odd-numbered terms of a sequence behave suf-

ficiently differently that it is desirable to investigate their convergence separately. The
following theorem, whose proof is omitted, is helpful for that purpose.
December 30, 2008 11:14 m91-ch09 9.1.4
Sheet theorem
number A sequence
7 Page number 602 converges to a limit
cyan magenta L ifblack
yellow and only if the sequences of
even-numbered terms and odd-numbered terms both converge to L.

Example 7 The sequence

1 1 1 1 1 1
, , , , , ,...
2 3 22 32 23 33
converges to 0, since the even-numbered terms and the odd-numbered terms both converge
to 0, and the sequence 1, 1 , 1, 1 , 1, 1 , . . .
{cn} 2 3 4
diverges, since the odd-numbered terms converge to 1 and the even-numbered terms con-
L
verge to 0.
{bn} {an}
THE SQUEEZING THEOREM FOR SEQUENCES
If an → L and cn → L, then The following theorem, illustrated in Figure 9.1.5, is an adaptation of the Squeezing The-
bn → L.
orem (1.6.4) to sequences. This theorem will be useful for finding limits of sequences that
Figure 9.1.5 cannot be obtained directly. The proof is omitted.

9.1.5 theorem (The Squeezing Theorem for Sequences) Let {an }, {bn }, and {cn } be se-
quences such that
an ≤ bn ≤ cn ( for all values of n beyond some index N )
If the sequences {an } and {cn } have a common limit L as n → +⬁, then {bn } also has
the limit L as n → +⬁.

9.1.6 theorem If lim |an | = 0, then lim an = 0.

n → +⬁ n → +⬁

Example 9 Consider the sequence

1 1 1 1
1, − , 2 , − 3 , . . . , (−1)n n , . . .
2 2 2 2
If we take the absolute value of each term, we obtain the sequence
1 1 1 1
, 2, 3,..., n,...
1,
2 2 2 2
which, as shown in Example 4, converges to 0. Thus, from Theorem 9.1.6 we have

1
lim (−1)n n = 0
n → +⬁ 2

EXERCISE SET 9.1 Graphing Utility

1. In each part, find a formula for the general term of the se-
quence, starting with n = 1. 7–22 Write out the first five terms of the sequence, determine
1 1 1 1 1 1 whether the sequence converges, and if so find its limit. ■
(a) 1, , , ,... (b) 1, − , , − , . . .
3 9 27 3 9 27
+⬁
1 3 5 7 1 4 9 16 n2 π +⬁
(c) , , , , . . . (d) √ , √ , √ , √ ,... 8. 12. n sin
2 4 6 8 π 3 π 4 π 5π 2n + 1 n=1 n n=1
19. {n2 e−n }+⬁
n=1
4. In each part, find a formula for the general term using fac-
torials and starting with n = 1.
(a) 1 · 2, 1 · 2 · 3 · 4, 1 · 2 · 3 · 4 · 5 · 6,
1 · 2 · 3 · 4 · 5 · 6 · 7 · 8, . . .
(b) 1, 1 · 2 · 3, 1 · 2 · 3 · 4 · 5, 1 · 2 · 3 · 4 · 5 · 6 · 7, . . .
December 30, 2008 11:14 m91-ch09 Sheet number 19 Page number 614 cyan magenta yellow black

9.3 INFINITE SERIES

SUMS OF INFINITE SERIES

Our first objective is to define what is meant by the “sum” of infinitely many real numbers.
We begin with some terminology.

9.3.1 definition An infinite series is an expression that can be written in the form

⬁
uk = u1 + u2 + u3 + · · · + uk + · · ·
k=1

The numbers u1 , u2 , u3 , . . . are called the terms of the series.

Since it is impossible to add infinitely many numbers together directly, sums of infinite
series are defined and computed by an indirect limiting process. To motivate the basic idea,
consider the decimal
0.3333 . . . (1)

This can be viewed as the infinite series

0.3 + 0.03 + 0.003 + 0.0003 + · · ·
or, equivalently,
3 3 3 3
+ + 3 + 4 + ··· (2)
10 102 10 10
Since (1) is the decimal expansion of 13 , any reasonable definition for the sum of an infinite
series should yield 13 for the sum of (2). To obtain such a definition, consider the following
sequence of (finite) sums:
3
s1 = = 0.3
10
3 3
s2 = + = 0.33
10 102
3 3 3
s3 = + 2 + 3 = 0.333
10 10 10
3 3 3 3
s4 = + 2 + 3 + 4 = 0.3333
10 10 10 10
.
.
The sequence of numbers s1 , s2 , s3 , s4 , . . . (Figure 9.3.1) can be viewed as a succession
of approximations to the “sum” of the infinite series, which we want to be 13 . As we
progress through the sequence, more and more terms of the infinite series are used, and the
approximations get better and better, suggesting that the desired sum of 13 might be the limit
of this sequence of approximations. To see that this is so, we must calculate the limit of the
general term in the sequence of approximations, namely,
3 3 3
sn = + 2 + ··· + n (3)
10 10 10
The problem of calculating
December 30, 2008 11:14 m91-ch09 Sheet number 20 Page number 615 cyan magenta yellow black

y 3 3 3
lim sn = lim + + ··· + n
0.4 n → +⬁ n → +⬁ 10 102 10
is complicated by the fact that both the last term and the number of terms in the sum change
1/3 with n. It is best to rewrite such limits in a closed form in which the number of terms does
0.3
not vary, if possible. (See the discussion of closed form and open form following Example
1
2 in Section 5.4.) To do this, we multiply both sides of (3) by 10 to obtain
1 3 3 3 3
sn = 2 + 3 + · · · + n + n+1 (4)
n 10 10 10 10 10
1 2 3 4 and then subtract (4) from (3) to obtain
1 3 3
0.3, 0.33, 0.333, ... sn − sn = − n+1
10 10 10
Figure 9.3.1
9 3 1
sn = 1− n
10 10 10
1 1
sn = 1− n
3 10
Since 1/10 n → 0 as n → +⬁, it follows that
1 1 1
lim sn = lim 1− n =
n → +⬁ n → +⬁ 3 10 3
.

which we denote by writing

1 3 3 3 3
= + + 3 + ··· + n + ···
3 10 102 10 10
Motivated by the preceding example, we are now ready to define the general concept of
the “sum” of an infinite series
u 1 + u 2 + u 3 + · · · + uk + · · ·
We begin with some terminology: Let sn denote the sum of the initial terms of the series,
up to and including the term with index n. Thus,
s1 = u1
s2 = u1 + u2
s3 = u1 + u2 + u3
..
.

n
sn = u1 + u2 + u3 + · · · + un = uk
k=1

The number sn is called the nth partial sum of the series and the sequence {sn }+ ⬁
n=1 is called
the sequence of partial sums.
As n increases, the partial sum sn = u1 + u2 + · · · + un includes more and more terms
of the series. Thus, if sn tends toward a limit as n → +⬁, it is reasonable to view this limit
as the sum of all the terms in the series. This suggests the following definition.
9.3.2 definition Let {sn } be the sequence of partial sums of the series
WARNING u1 + u2 + u3 + · · · + uk + · · ·
In everyday language the words “se- If the sequence {sn } converges to a limit S, then the series is said to converge to S, and
quence” and “series” are often used S is called the sum of the series. We denote this by writing
interchangeably. However, in mathe-
matics there is a distinction between 
⬁
these two words—a sequence is a suc- S= uk
cession whereas a series is a sum. It is k=1
essential that you keep this distinction
in mind.
If the sequence of partial sums diverges, then the series is said to diverge. A divergent
series has no sum.
December 30, 2008 11:14 m91-ch09 Sheet number 22 Page number 617 cyan magenta yellow black

Example 1 Determine whether the series

1 − 1 + 1 − 1 + 1 − 1 + ···
converges or diverges. If it converges, find the sum.

Solution. It is tempting to conclude that the sum of the series is zero by arguing that the
positive and negative terms cancel one another. However, this is not correct; the problem
is that algebraic operations that hold for finite sums do not carry over to infinite series in
all cases. Later, we will discuss conditions under which familiar algebraic operations can
be applied to infinite series, but for this example we turn directly to Definition 9.3.2. The
partial sums are
s1 = 1
s2 = 1 − 1 = 0
s3 = 1 − 1 + 1 = 1
s4 = 1 − 1 + 1 − 1 = 0
y
and so forth. Thus, the sequence of partial sums is
1.5
1, 0, 1, 0, 1, 0, . . .
1
(Figure 9.3.2). Since this is a divergent sequence, the given series diverges and consequently
0.5 has no sum.
n
2 4 6 8 10
GEOMETRIC SERIES
1, 0, 1, 0, 1, 0, ... In many important series, each term is obtained by multiplying the preceding term by some
fixed constant. Thus, if the initial term of the series is a and each term is obtained by
Figure 9.3.2
multiplying the preceding term by r, then the series has the form

⬁
ar k = a + ar + ar 2 + ar 3 + · · · + ar k + · · · (a = 0) (5)
k=0

Such series are called geometric series, and the number r is called the ratio for the series.
Here are some examples:
1 + 2 + 4 + 8 + · · · + 2k + · · · a = 1, r = 2

3 3 3 3
+ + 3 + ··· + k + ··· a= 3
10 , r = 1
10
10 102 10 10
1 1 1 1 1
− + − + · · · + (−1)k+1 k + · · · a = 12 , r = − 12
2 4 8 16 2
1 + 1 + 1 + ··· + 1 + ··· a = 1, r = 1

1 − 1 + 1 − 1 + · · · + (−1)k+1 + · · · a = 1, r = −1

1 + x + x2 + x3 + · · · + xk + · · · a = 1, r = x

The following theorem is the fundamental result on convergence of geometric series.

Sometimes it is desirable to start the in-
dex of summation of an infinite series
at k = 0 rather than k = 1, in which
case we would call u0 the zeroth term
9.3.3 theorem A geometric series
and s0 = u0 the zeroth partial sum. 
⬁
One can prove that changing the start- ar k = a + ar + ar 2 + · · · + ar k + · · · (a = 0)
ing value for the index of summation k=0
of an infinite series has no effect on
the convergence, the divergence, or the converges if |r| < 1 and diverges if |r| ≥ 1. If the series converges, then the sum is
sum. If we had started the index at
k = 1 in (5), then the series would be 
⬁
a
expressed as ar k =
k=0
1−r
⬁

ar k−1
k=1

Since this expression is more compli-

cated than (5), we started the index at
k = 0.
December 30, 2008 11:14 m91-ch09 Sheet number 23 Page number 618 cyan magenta yellow black

y Example 2 In each part, determine whether the series converges, and if so find its
20/ 3 sum. ⬁
5 ⬁
6 (a) k
(b) 32k 51−k
5 k=0
4 k=1

n Solution (a). This is a geometric series with a = 5 and r = 14 . Since |r| = 1

4
< 1, the
1 2 3 4 5 6 7 8 series converges and the sum is
∞ a 5 20
Partial sums for
5 = =
k=0 4k 1−r 1− 1
4
3
Figure 9.3.3 (Figure 9.3.3).

Solution (b). This is a geometric series in concealed form, since we can rewrite it as

⬁ ⬁
9k ⬁
9 k−1
32k 51−k = k−1
= 9
k=1 k=1
5 k=1
5

Since r = 9
5
> 1, the series diverges.

Example 3 Find the rational number represented by the repeating decimal

0.784784784 . . .
T E C H N O L O GY M A ST E R Y
Solution. We can write
Computer algebra systems have com-
mands for finding sums of convergent 0.784784784 . . . = 0.784 + 0.000784 + 0.000000784 + · · ·
series. If you have a CAS, use it to com-
pute the sums in Examples 2 and 3. so the given decimal is the sum of a geometric series with a = 0.784 and r = 0.001. Thus,
a 0.784 0.784 784
0.784784784 . . . = = = =
1−r 1 − 0.001 0.999 999

Example 4 In each part, find all values of x for which the series converges, and find
the sum of the series for those values of x.
 ⬁
3x 3x 2 3x 3 3(−1)k k
(a) xk (b) 3 − + − + ··· + x + ···
k=0
2 4 8 2k

Solution (a). The expanded form of the series is


⬁
xk = 1 + x + x2 + · · · + xk + · · ·
k=0
The series is a geometric series with a = 1 and r = x, so it converges if |x| < 1 and diverges
otherwise. When the series converges its sum is
 ⬁
1
xk =
k=0
1 − x

Solution (b). This is a geometric series with a = 3 and r = −x /2. It converges if

| −x /2| < 1, or equivalently, when |x| < 2. When the series converges its sum is
⬁  x k 3 6
3 − =  x =
k=0
2 1− − 2+x
2
December 30, 2008 11:14 m91-ch09 Sheet number 24 Page number 619 cyan magenta yellow black

TELESCOPING SUMS

Example 5 Determine whether the series

⬁
1 1 1 1 1
= + + + + ···
k=1
k(k + 1) 1 · 2 2 · 3 3 · 4 4 ·5
converges or diverges. If it converges, find the sum.

Solution. The nth partial sum of the series is


n
1 1 1 1 1
sn = = + + + ··· +
k=1
k(k + 1) 1·2 2·3 3·4 n(n + 1)
We will begin by rewriting sn in closed form. This can be accomplished by using the method
of partial fractions to obtain (verify)
1 1 1
= −
k(k + 1) k k+1
from which we obtain the sum
n
1 1
sn = −
k=1
k k + 1

1 1 1 1 1 1 1
The sum in (10) is an example of a = 1− + − + − + ··· + −
telescoping sum. The name is derived 2 2 3 3 4 n n+1
from the fact that in simplifying the
sum, one term in each parenthetical ex- 1 1 1 1 1 1 1
pression cancels one term in the next =1+ − + + − + + ··· + − + −
2 2 3 3 n n n+1
parenthetical expression, until the en-
tire sum collapses (like a folding tele-
1
scope) into just two terms. =1− (10)
n+1

Thus, 
⬁
1 1
= lim sn = lim 1− =1
k=1
k(k + 1) n → +⬁ n → +⬁ n+1

HARMONIC SERIES
One of the most important of all diverging series is the harmonic series,


⬁
1 1 1 1 1
=1+ + + + + ···
k=1
k 2 3 4 5

EXERCISE SET 9.3 C CAS

1–2 In each part, find exact values for the first four partial 3–14 Determine whether the series converges, and if so find its
sums, find a closed form for the nth partial sum, and determine sum. ■
whether the series converges by calculating the limit of the nth
partial sum. If the series converges, then state its sum. ■ 
⬁
3 k+1 
⬁
1 
⬁
6. − 9. 14. 53k 71−k
2 9k 2 + 3k − 2
⬁
1 k 
⬁ 
⬁
1 1 k=1 k=1 k=1
2. (a) (b) 4k−1 (c) −
k=1
4 k=1 k=1
k+3 k+4
December 30, 2008 11:14 m91-ch09 Sheet number 28 Page number 623 cyan magenta yellow black
December 30, 2008 11:14 m91-ch09 Sheet number 29 Page number 624 cyan magenta yellow black

9.4 CONVERGENCE TESTS

THE DIVERGENCE TEST

In stating general
 results about convergence or divergence of series, it is convenient to use
the notation uk as a generic notation for a series, thus avoiding the issue of whether the
sum begins with k = 0 or k = 1 or some other value. Indeed, we will see shortly that the
startingindex value is irrelevant to the issue of convergence. The kth term in an infinite
series uk is called the general term of the series. The following theorem establishes
a relationship between the limit of the general term and the convergence properties of a
series.

9.4.1 theorem (The Divergence Test)


(a) If lim uk = 0, then the series uk diverges.
k → +⬁

(b) If lim uk = 0, then the series uk may either converge or diverge.
k → +⬁

WARNING The alternative form of part (a) given in the preceding proof is sufficiently important
that we state it separately for future reference.
The converse of Theorem 9.4.2 is false;
i.e., showing that

lim uk = 0 
k → +⬁ 9.4.2 theorem If the series uk converges, then lim uk = 0.
 k → +⬁
does not prove that uk converges,
since this property may hold for diver-
gent as well as convergent series. This
is illustrated in the proof of part (b) of
Theorem 9.4.1.
Example 1 The series

⬁
k 1 2 3 k
= + + + ··· + + ···
k=1
k+1 2 3 4 k+1

diverges since
k 1
lim = lim =1=0
k → +⬁ k+1 k → +⬁ 1 + 1/k
December 30, 2008 11:14 m91-ch09 Sheet number 30 Page number 625 cyan magenta yellow black

ALGEBRAIC PROPERTIES OF INFINITE SERIES

For brevity, the proof of the following result is omitted.

9.4.3 theorem
   
See Exercises 27 and 28 for an explo-
(a) If uk and vk are convergent series, then (uk + vk ) and (uk − vk ) are
 convergent series and the sums of these series are related by
ration of what happens when uk or

vk diverge.

⬁ 
⬁ 
⬁
(uk + vk ) = uk + vk
k=1 k=1 k=1
⬁ ⬁ ⬁
(uk − vk ) = uk − vk
k=1 k=1 k=1
 
(b) If c is a nonzero constant, then the series uk and cuk both converge or both
diverge. In the case of convergence, the sums are related by

⬁ 
⬁
cuk = c uk
k=1 k=1

(c) Convergence or divergence is unaffected by deleting a finite number of terms from

a series; in particular, for any positive integer K, the series
WARNING

⬁
uk = u1 + u2 + u3 + · · ·
Do not read too much into part (c)
k=1
of Theorem 9.4.3. Although conver-
gence is not affected when finitely ⬁
many terms are deleted from the be- uk = uK + uK+1 + uK+2 + · · ·
ginning of a convergent series, the sum k=K
of the series is changed by the removal
of those terms. both converge or both diverge.

Example 2 Find the sum of the series


⬁
3 2
k
− k−1
k=1
4 5

Solution. The series

⬁
3 3 3 3
k
= + 2 + 3 + ···
k=1
4 4 4 4
 
is a convergent geometric series a = 34 , r = 14 , and the series
⬁
2 2 2 2
k−1
= 2 + + 2 + 3 + ···
k=1
5 5 5 5
 
is also a convergent geometric series a = 2, r = 15 . Thus, from Theorems 9.4.3(a) and
9.3.3 the given series converges and

⬁
3 2 ⬁
3 ⬁
2
k
− k−1 = k
− k−1
k=1
4 5 k=1
4 k=1
5
3
2 3
= 4
− =−
1− 1
4
1− 1
5
2
December 30, 2008 11:14 m91-ch09 Sheet number 31 Page number 626 cyan magenta yellow black

Example 3 Determine whether the following series converge or diverge.


⬁
5 5 5 5 ⬁
1 1 1 1
(a) =5+ + + ··· + + ··· (b) = + + + ···
k=1
k 2 3 k k=10
k 10 11 12

Solution. The first series is a constant times the divergent harmonic series, and hence
diverges by part (b) of Theorem 9.4.3. The second series results by deleting the first nine
terms from the divergent harmonic series, and hence diverges by part (c) of Theorem 9.4.3.

THE INTEGRAL TEST


9.4.4 theorem (The Integral Test) Let uk be a series with positive terms. If f is
a function that is decreasing and continuous on an interval [a, +⬁) and such that
uk = f(k) for all k ≥ a, then
⬁  +⬁
uk and f(x) dx
k=1 a

both converge or both diverge.

Example 4 Show that the integral test applies, and use the integral test to determine
whether the following series converge or diverge.

⬁
1 ⬁
1
(a) (b)
k=1
k k=1
k 2

Solution (a). We already know that this is the divergent harmonic series, so the integral
test will simply illustrate another way of establishing the divergence.
Note first that the series has positive terms, so the integral test is applicable. If we replace
k by x in the general term 1/k, we obtain the function f(x) = 1/x, which is decreasing and
continuous for x ≥ 1 (as required to apply the integral test with a = 1). Since
 +⬁  b
1 1
dx = lim dx = lim [ln b − ln 1] = +⬁
1 x b → +⬁ 1 x b → +⬁

the integral diverges and consequently so does the series.

WARNING Solution (b). Note first that the series has positive terms, so the integral test is applicable.
If we replace k by x in the general term 1/k 2 , we obtain the function f(x) = 1/x 2 , which
In part (b) of Example 4, do not erro-
neously conclude that the sum of the
is decreasing and continuous for x ≥ 1. Since
 +⬁  b  
series is 1 because the value of the cor-
1 dx 1 b 1
responding integral is 1. You can see dx = lim = lim − = lim 1 − =1
that this is not so since the sum of the 1 x2 b → +⬁ 1 x 2 b → +⬁ x 1 b → +⬁ b
first two terms alone exceeds 1. Later,
the integral converges and consequently the series converges by the integral test with a = 1.
we will see that the sum of the series
is actually π2 /6.
December 30, 2008 11:14 m91-ch09 Sheet number 32 Page number 627 cyan magenta yellow black

p-SERIES
The series in Example 4 are special cases of a class of series called p-series or hyperhar-
monic series. A p-series is an infinite series of the form
⬁
1 1 1 1
p
= 1 + p + p + ··· + p + ···
k=1
k 2 3 k

where p > 0. Examples of p-series are


⬁
1 1 1 1
=1+ + + ··· + + ··· p=1
k=1
k 2 3 k
⬁
1 1 1 1
= 1 + 2 + 2 + ··· + 2 + ··· p=2
k=1
k2 2 3 k
⬁
1 1 1 1
√ = 1 + √ + √ + ··· + √ + ··· p= 1
2
k=1
k 2 3 k
The following theorem tells when a p-series converges.

9.4.5 theorem (Convergence of p-Series)

⬁
1 1 1 1
p
= 1 + p + p + ··· + p + ···
k=1
k 2 3 k

converges if p > 1 and diverges if 0 < p ≤ 1.

Example 5 1 1 1
1+ √
3
+ √
3
+ ··· + √
3
+ ···
2 3 k
diverges since it is a p-series with p = 1
3
< 1.
EXERCISE SET 9.4 Graphing Utility C CAS

1. Use Theorem 9.4.3 to find the sum of each series. 7–8 Confirm that the integral test is applicable and use it to
1 1 1 1 1 1 determine whether the series converges. ■
(a) + + + 2 + ··· + + k + ···
2 4 22 4 2k 4
⬁
1 1 
⬁
(b) − k
5 k k(k + 1) 8. (a)
k=1 k=1
1 + k2
3–4 For each given p-series, identify p and determine whether 9–24 Determine whether the series converges. ■
the series converges. ■

⬁
k2 + 1 
⬁
1

⬁ 
⬁ 
⬁ 
⬁ 18. 20. √
1 1 1 k2 + 3 k +1
2
4. (a) k −4/3 (b) √ (c) √ (d) k=1 k=1
k=1 k=1
4
k k=1
3 5
k k=1
kπ
December 30, 2008 11:14 m91-ch09 Sheet number 36 Page number 631 cyan magenta yellow black
December 30, 2008 11:14 m91-ch09 Sheet number 37 Page number 632 cyan magenta yellow black

9.5 THE COMPARISON, RATIO, AND ROOT TESTS

THE COMPARISON TEST
We will begin with a test that is useful in its own right and is also the building block for other
important convergence tests. The underlying idea of this test is to use the known conver-
gence or divergence of a series to deduce the convergence or divergence of another series.
 
9.5.1 theorem (The Comparison Test) Let ⬁k=1 ak and ⬁k=1 bk be series with non-
negative terms and suppose that

It is not essential in Theorem 9.5.1 that

a1 ≤ b1 , a2 ≤ b2 , a3 ≤ b3 , . . . , ak ≤ bk , . . .
the condition ak ≤ bk hold for all k , as
(a) If the “bigger series” #bk converges, then the “smaller series” #ak also con-
stated; the conclusions of the theorem
remain true if this condition is eventu-
verges.
ally true.
(b) If the “smaller series” #ak diverges, then the “bigger series” #bk also diverges.

USING THE COMPARISON TEST

There
 are two steps required for using the comparison test to determine whether a series
uk with positive terms converges:


Step 1. Guess at whether the series uk converges or diverges.

Step 2. Find a series that proves the guess to be correct. That is, if we guess that uk
diverges, we must find adivergent series whose terms  are “smaller” than the
corresponding terms of uk , and if we guess that uk converges, we must
finda convergent series whose terms are “bigger” than the corresponding terms
of uk .

9.5.2 informal principle Constant terms in the denominator of uk can usually

be deleted without affecting the convergence or divergence of the series.

9.5.3 informal principle If a polynomial in k appears as a factor in the numer-

ator or denominator of uk , all but the leading term in the polynomial can usually be
discarded without affecting the convergence or divergence of the series.

Example 1 Use the comparison test to determine whether the following series con-
verge or diverge. ⬁ ⬁
1 1
(a) √ (b) 2+k
k=1
k− 2
1
k=1
2k

Solution (a). According to Principle 9.5.2, we should be able to drop the constant in
the denominator without affecting the convergence or divergence. Thus, the given series is
likely to behave like ⬁
1
√ (1)
k=1
k
 
which is a divergent p-series p = 12 . Thus, we will guess that the given series diverges
and try to prove this by finding a divergent series that is “smaller” than the given series.
However, series (1) does the trick since
1 1
√ >√ for k = 1, 2, . . .
k− 1
2
k
Thus, we have proved that the given series diverges.
December 30, 2008 11:14 m91-ch09 Sheet number 38 Page number 633 cyan magenta yellow black

Solution (b). According to Principle 9.5.3, we should be able to discard all but the leading
term in the polynomial without affecting the convergence or divergence. Thus, the given
series is likely to behave like  1 1
⬁ ⬁
1
= (2)
k=1
2k 2 2 k=1 k 2

which converges since it is a constant times a convergent p-series (p = 2). Thus, we will
guess that the given series converges and try to prove this by finding a convergent series
that is “bigger” than the given series. However, series (2) does the trick since
1 1
< 2 for k = 1, 2, . . .
2k 2+k 2k
Thus, we have proved that the given series converges.

THE LIMIT COMPARISON TEST

In the last example, Principles 9.5.2 and 9.5.3 provided the guess about convergence or
divergence as well as the series needed to apply the comparison test. Unfortunately, it is
not always so straightforward to find the series required for comparison, so we will now
consider an alternative to the comparison test that is usually easier to apply. The proof is
given in Appendix D.

 
9.5.4 theorem (The Limit Comparison Test) Let ak and bk be series with positive
terms and suppose that ak
ρ = lim
k → +⬁ bk

If ρ is finite and ρ > 0, then the series both converge or both diverge.

The cases where ρ = 0 or ρ = +⬁ are discussed in the exercises (Exercise 54).

To use the limit comparison test we
 must again first guess at the convergence or diver-
gence of ak and then find a series bk that supports our guess. The following example
illustrates this principle.

Example 2 Use the limit comparison test to determine whether the following series
converge or diverge.

⬁
1 
⬁
1 
⬁
3k 3 − 2k 2 + 4
(a) √ (b) (c)
k=1
k+1 k=1
2k 2 + k k=1
k7 − k3 + 2

Solution (a). As in Example 1, Principle 9.5.2 suggests that the series is likely to behave
like the divergent p-series (1). To prove that the given series diverges, we will apply the
limit comparison test with
1 1
ak = √ and bk = √
k+1 k
We obtain
√
ak k 1
ρ = lim = lim √ = lim =1
k → +⬁ bk k → +⬁ k + 1 k → +⬁ 1+ √
1
k
Since ρ is finite and positive, it follows from Theorem 9.5.4 that the given series diverges.
December 30, 2008 11:14 m91-ch09 Sheet number 39 Page number 634 cyan magenta yellow black

Solution (b). As in Example 1, Principle 9.5.3 suggests that the series is likely to behave
like the convergent series (2). To prove that the given series converges, we will apply the
limit comparison test with
1 1
ak = and bk =
2k 2 + k 2k 2
We obtain ak 2k 2 2
ρ = lim = lim = lim =1
k → +⬁ bk k → +⬁ 2k 2 + k k → +⬁ 1
2+
k
Since ρ is finite and positive, it follows from Theorem 9.5.4 that the given series converges,
which agrees with the conclusion reached in Example 1 using the comparison test.

Solution (c). From Principle 9.5.3, the series is likely to behave like

⬁
3k 3 ⬁
3
= (3)
k=1
k7 k=1
k 4

which converges since it is a constant times a convergent p-series. Thus, the given series
is likely to converge. To prove this, we will apply the limit comparison test to series (3)
and the given series. We obtain
3k 3 − 2k 2 + 4
k 7 − k 3 + 2 = lim 3k − 2k + 4k = 1
7 6 4
ρ = lim
k → +⬁ 3 k → +⬁ 3k 7 − 3k 3 + 6
k 4

Since ρ is finite and nonzero, it follows from Theorem 9.5.4 that the given series converges,
since (3) converges.

THE RATIO TEST


9.5.5 theorem (The Ratio Test) Let uk be a series with positive terms and suppose
that uk+1
ρ = lim
k → +⬁ uk

(a) If ρ < 1, the series converges.

(b) If ρ > 1 or ρ = +⬁, the series diverges.
(c) If ρ = 1, the series may converge or diverge, so that another test must be tried.

Example 3 Each of the following series has positive terms, so the ratio test applies. In
each part, use the ratio test to determine whether the following series converge or diverge.
⬁
1 ⬁
k 
⬁
kk 
⬁
(2k)! 
⬁
1
(a) (b) (c) (d) (e)
k=1
k! k=1
2 k
k=1
k! k=3
4k k=1
2k − 1
December 30, 2008 11:14 m91-ch09 Sheet number 40 Page number 635 cyan magenta yellow black

Solution (a). The series converges, since

uk+1 1/(k + 1)! k! 1
ρ = lim = lim = lim = lim =0<1
k → +⬁ uk k → +⬁ /
1 k! k → +⬁ (k + 1)! k → +⬁ k+1

Solution (b). The series converges, since

uk+1 k + 1 2k 1 k+1 1
ρ = lim = lim k+1 · = lim = <1
k → +⬁ uk k → +⬁ 2 k 2 k → +⬁ k 2

Solution (c). The series diverges, since

k
uk+1 (k + 1)k+1 k! (k + 1)k 1
ρ = lim = lim · k = lim = lim 1 + =e>1
k → +⬁ uk k → +⬁ (k + 1)! k k → +⬁ kk k → +⬁ k
See Formula (7)
of Section 1.3

Solution (d). The series diverges, since

uk+1 [2(k + 1)]! 4k (2k + 2)! 1
ρ = lim = lim · = lim ·
k → +⬁ uk k → +⬁ 4k+1 (2k)! k → +⬁ (2k)! 4
(2k + 2)(2k + 1)(2k)! 1 1
= lim · = lim (2k + 2)(2k + 1) = +⬁
k → +⬁ (2k)! 4 4 k → +⬁

Solution (e). The ratio test is of no help since

uk+1 1 2k − 1 2k − 1
ρ = lim = lim · = lim =1
k → +⬁ uk k → +⬁ 2(k + 1) − 1 1 k → +⬁ 2k + 1

However, the integral test proves that the series diverges since
 +⬁  b b
dx dx 1
= lim = lim ln(2x − 1) = +⬁
1 2x − 1 b → +⬁ 1 2x − 1 b → +⬁ 2 1
Both the comparison test and the limit comparison test would also have worked here (verify).

THE ROOT TEST

In cases where it is difficult or inconvenient to find the limit required for the ratio test, the
next test is sometimes useful. Since its proof is similar to the proof of the ratio test, we will
omit it.


9.5.6 theorem (The Root Test) Letuk be a series with positive terms and suppose
that √
ρ = lim k uk = lim (uk )1/k
k → +⬁ k → +⬁

(a) If ρ < 1, the series converges.

(b) If ρ > 1 or ρ = +⬁, the series diverges.
(c) If ρ = 1, the series may converge or diverge, so that another test must be tried.

Example 4 Use the root test to determine whether the following series converge or
diverge. ⬁ ⬁
4k − 5 k 1
(a) (b)
k=2
2k + 1 k=1
(ln(k + 1))k
December 30, 2008 11:14 m91-ch09 Sheet number 41 Page number 636 cyan magenta yellow black

Solution (a). The series diverges, since

4k − 5
ρ = lim (uk )1/k = lim =2>1
k → +⬁ k → +⬁ 2k + 1

Solution (b). The series converges, since

1
ρ = lim (uk )1/k = lim =0<1
k → +⬁ k → +⬁ ln(k + 1)

EXERCISE SET 9.5

1–2 Make a guess about the convergence or divergence of the 11–16 Use the ratio test to determine whether the series con-
series, and confirm your guess using the comparison test. ■ verges. If the test is inconclusive, then say so. ■
⬁
3k ⬁
11. k
k! 16. 2+1
k=1
k=1
k
⬁
k+1 
⬁
2 ⬁
2. (a) (b) 1 k
k 2−k k4 + k 14. k
k=2 k=1
k=1
2

5–10 Use the limit comparison test to determine whether the 17–20 Use the root test to determine whether the series con-
series converges. ■ verges. If the test is inconclusive, then say so. ■
⬁ ⬁
k k
1 18.
10. 100
k=1
(2k + 3)17 k=1

⬁
20. (1 − e−k )k
k=1
December 30, 2008 11:14 m91-ch09 Sheet number 43 Page number 638 cyan magenta yellow black

9.6 ALTERNATING SERIES; ABSOLUTE CONVERGENCE

ALTERNATING SERIES
Series whose terms alternate between positive and negative, called alternating series, are
of special importance. Some examples are

⬁
1 1 1 1 1
(−1)k+1 = 1 − + − + − ···
k=1
k 2 3 4 5

⬁
1 1 1 1 1
(−1)k = −1 + − + − + · · ·
k=1
k 2 3 4 5

In general, an alternating series has one of the following two forms:


⬁
(−1)k+1 ak = a1 − a2 + a3 − a4 + · · · (1)
k=1


⬁
(−1)k ak = −a1 + a2 − a3 + a4 − · · · (2)
k=1

where the ak ’s are assumed to be positive in both cases.

The following theorem is the key result on convergence of alternating series.

9.6.1 theorem (Alternating Series Test) An alternating series of either form (1) or form
(2) converges if the following two conditions are satisfied:
(a) a1 ≥ a2 ≥ a3 ≥ · · · ≥ ak ≥ · · ·
(b) lim ak = 0
k → +⬁

Example 1 Use the alternating series test to show that the following series converge.

⬁
1 
⬁
k+3
(a) (−1)k+1 (b) (−1)k+1
k=1
k k=1
k(k + 1)

Solution (a). The two conditions in the alternating series test are satisfied since
The series in part (a) of Example 1
is called the alternating harmonic 1 1 1
ak = > = ak+1 and lim ak = lim =0
series. Note that this series converges, k k+1 k → +⬁ k → +⬁ k
whereas the harmonic series diverges.

Solution (b). The two conditions in the alternating series test are satisfied since
ak+1 k+4 k(k + 1) k 2 + 4k k 2 + 4k
= · = 2 = 2 <1
ak (k + 1)(k + 2) k+3 k + 5k + 6 (k + 4k) + (k + 6)
so
ak > ak+1

and 1 3
k+3 + 2
lim ak = lim = lim k k =0
k → +⬁ k → +⬁ k(k + 1) k → +⬁ 1
1+
k
December 30, 2008 11:14 m91-ch09 Sheet number 46 Page number 641 cyan magenta yellow black

ABSOLUTE CONVERGENCE
The series 1 1 1 1 1 1
1 − − 2 + 3 + 4 − 5 − 6 + ···
2 2 2 2 2 2
does not fit in any of the categories studied so far—it has mixed signs but is not alternating.
We will now develop some convergence tests that can be applied to such series.

9.6.3 definition A series


⬁
uk = u1 + u2 + · · · + uk + · · ·
k=1

is said to converge absolutely if the series of absolute values


⬁
|uk | = |u1 | + |u2 | + · · · + |uk | + · · ·
k=1

converges and is said to diverge absolutely if the series of absolute values diverges.

Example 3 Determine whether the following series converge absolutely.

1 1 1 1 1 1 1 1 1
(a) 1 − − + 3 + 4 − 5 − ··· (b) 1 − + − + − ···
2 22 2 2 2 2 3 4 5

Solution (a). The series of absolute values is the convergent geometric series
1 1 1 1 1
1+ + 2 + 3 + 4 + 5 + ···
2 2 2 2 2
so the given series converges absolutely.

Solution (b). The series of absolute values is the divergent harmonic series
1 1 1 1
1+ + + + + ···
2 3 4 5
so the given series diverges absolutely.

It is important to distinguish between the notions of convergence and absolute conver-

gence. For example, the series in part (b) of Example 3 converges, since it is the alternating
harmonic series, yet we demonstrated that it does not converge absolutely. However, the
following theorem shows that if a series converges absolutely, then it converges.

9.6.4 theorem If the series

Theorem 9.6.4 provides a way of infer- 
⬁
ring convergence of a series with posi- |uk | = |u1 | + |u2 | + · · · + |uk | + · · ·
tive and negative terms from a related k=1
series with nonnegative terms (the se-
ries of absolute values). This is impor- converges, then so does the series
tant because most of the convergence
tests that we have developed apply 
⬁
only to series with nonnegative terms. uk = u1 + u2 + · · · + uk + · · ·
k=1
December 30, 2008 11:14 m91-ch09 Sheet number 48 Page number 643 cyan magenta yellow black

y
1
0.8
0.6
{sn}
0.4
0.2
n
2 4 6 8 10
−0.2
−0.4 Example 4 Show that the following series converge.
{uk}
−0.6
1 1 1 1 1 1 
⬁
cos k
(a) (a) 1 − − 2 + 3 + 4 − 5 − 6 + ··· (b)
2 2 2 2 2 2 k=1
k2
y
1
0.8 Solution (a). Observe that this is not an alternating series because the signs alternate in
0.6 pairs after the first term. Thus, we have no convergence test that can be applied directly.
0.4
{sn} However, we showed in Example 3(a) that the series converges absolutely, so Theorem
0.2
n
9.6.4 implies that it converges (Figure 9.6.4a).
4 6 8 10
−0.2 Solution (b). With the help of a calculating utility, you will be able to verify that the
−0.4 {uk} signs of the terms in this series vary irregularly. Thus, we will test for absolute convergence.
−0.6 The series of absolute values is ⬁  
 cos k 
(b)  
 k2 
k=1
Graphs of the sequences of
terms and n th partial sums
However,  
for the series in Example 4  cos k  1
 
Figure 9.6.4
 k2  ≤ k2


But 1/k 2 is a convergent p-series (p = 2), so the series of absolute values converges
by the comparison test. Thus, the given series converges absolutely and hence converges
(Figure 9.6.4b).

THE RATIO TEST FOR ABSOLUTE CONVERGENCE

Although one cannot generally infer convergence or divergence of a series from absolute
divergence, the following variation of the ratio test provides a way of deducing divergence
from absolute divergence in certain situations. We omit the proof.


9.6.5 theorem (Ratio Test for Absolute Convergence) Let uk be a series with nonzero
terms and suppose that |uk+1 |
ρ = lim
k → +⬁ |uk |

(a) If ρ < 1, then the series uk converges absolutely and therefore converges.

(b) If ρ > 1 or if ρ = +⬁, then the series uk diverges.
(c) If ρ = 1, no conclusion about convergence or absolute convergence can be drawn
from this test.
December 30, 2008 11:14 m91-ch09 Sheet number 49 Page number 644 cyan magenta yellow black

Example 6 Use the ratio test for absolute convergence to determine whether the series
converges. ⬁ ⬁
2k (2k − 1)!
(a) (−1)k (b) (−1)k
k=1
k! k=1
3k

Solution (a). Taking the absolute value of the general term uk we obtain
 
 k
k2  2k

|uk | = (−1) =
k!  k!
Thus,
|uk+1 | 2k+1 k! 2
ρ = lim = lim · = lim =0<1
k → +⬁ |uk | k → +⬁ (k + 1)! 2k k → +⬁ k + 1

which implies that the series converges absolutely and therefore converges.

Solution (b). Taking the absolute value of the general term uk we obtain
 
 (2k − 1)!  (2k − 1)!
|uk | = (−1)k =
3k 3k
Thus, |uk+1 | [2(k + 1) − 1]! 3k
ρ = lim = lim ·
k → +⬁ |uk | k → +⬁ 3 k+1 (2k − 1)!
1 (2k + 1)! 1
= lim · = lim (2k)(2k + 1) = +⬁
k → +⬁ 3 (2k − 1)! 3 k → +⬁
which implies that the series diverges.

EXERCISE SET 9.6 C CAS

1–2 Show that the series converges by confirming that it satis-

fies the hypotheses of the alternating series test (Theorem 9.6.1).
■

⬁
(−1)k+1 
⬁
k
1. 2. (−1)k+1 k
k=1
2k + 1 k=1
3
December 30, 2008 11:14 m91-ch09 Sheet number 50 Page number 645 cyan magenta yellow black

Summary of Convergence Tests

name statement comments

Divergence Test If lim uk ≠ 0, then uk diverges. If lim uk = 0, then uk may or

k→+ ∞
(9.4.1) k→+ ∞ may not converge.

Let uk be a series with positive terms. If f is a function

that is decreasing and continuous on an interval [a, + ∞) This test only applies to series that
and such that uk = f (k) for all k ≥ a, then have positive terms.
Integral Test
∞ +∞
(9.4.4) uk and f (x) dx Try this test when f (x) is easy to
k =1 a integrate.
both converge or both diverge.
∞ ∞
Let a and
k=1 k
b
k=1 k
be series with nonnegative
This test only applies to series with
terms such that nonnegative terms.
Comparison Test
a1 ≤ b1, a2 ≤ b2, . . . , ak ≤ bk , . . .
(9.5.1)
Try this test as a last resort; other
If bk converges, then ak converges, and if ak
tests are often easier to apply.
diverges, then bk diverges.

Let ak and bk be series with positive terms and let This is easier to apply than the
Limit Comparison Test a comparison test, but still requires
r = lim k some skill in choosing the series
(9.5.4) k→+ ∞ bk
bk for comparison.
If 0 < r < + ∞, then both series converge or both diverge.

Let uk be a series with positive terms and suppose that

uk+1
r= lim uk
Ratio Test k→+ ∞ Try this test when uk involves
(9.5.5) factorials or kth powers.
(a) Series converges if r < 1.
(b) Series diverges if r > 1 or r = + ∞.
(c) The test is inconclusive if r = 1.

Let uk be a series with positive terms and suppose that

k
r= lim uk
Root Test k→ + ∞ Try this test when uk involves kth
(9.5.6) powers.
(a) The series converges if r < 1.
(b) The series diverges if r > 1 or r = + ∞.
(c) The test is inconclusive if r = 1.

If ak > 0 for k = 1, 2, 3, . . . , then the series

a1 − a 2 + a 3 − a 4 + . . .
Alternating Series Test −a1 + a2 – a3 + a4 − . . . This test applies only to alternating
(9.6.1) converge if the following conditions hold: series.
(a) a1 ≥ a2 ≥ a3 ≥ . . .
(b) lim ak = 0
k→+ ∞

Let uk be a series with nonzero terms and suppose that

| uk+1 |
Ratio Test for r = lim The series need not have positive
k→ + ∞ | uk |
Absolute Convergence terms and need not be alternating
(9.6.5) (a) The series converges absolutely if r < 1. to use this test.
(b) The series diverges if r > 1 or r = + ∞.
(c) The test is inconclusive if r = 1.