M ATH 140 A - HW 5 S OLUTIONS

Problem 1 (WR Ch 3 #8). If

P
a n b n converges.

a n converges, and if {b n } is monotonic and bounded, prove that

Solution. Theorem 3.42 states that if

(a) the partial sums of

a n form a bounded sequence;

(b) b 0 b 1 b 2 ;
(c) lim b n = 0,
n

then

a n b n converges.

First of all, since

a n converges, that means the sequence of partial sums {

Pk

n=1 a n }

is a con-

vergent sequence, so by Theorem 3.2(c) it is bounded, and thus part (a) is satisfied.
The problem with using this theorem with {b n } is that it doesnt necessarily converge to 0.
However, we can create a new sequence {c n } based on {b n } which has the properties we need to
apply the theorem. Since {b n } is monotonic and bounded, it converges by Theorem 3.14. Let
b = lim b n . Now we have two cases:
b 0 b Then b n must be decreasing, so define the sequence c n = b n b, which is monotonic decreasing as well, so it satisfies (b). Also, notice that
lim c n = lim (b n b) = ( lim b n ) b = b b = 0,

so it satisfies (c). Therefore,

X

an cn =

a n c n converges. Then we note that

an bn

an b

an bn =

an cn + b

an ,

and the right side of the equation consists of two convergent sequences, so

a n b n con-

verges.
b 0 b Then b n must be increasing, so let c n = b b n , which is monotonic decreasing, so it satisfies
P
(b). Once again, lim c n = 0, so it satisfies (c), so a n c n converges. Then
n

X
so

an cn =

an b

an bn

an bn = b

an

an cn ,

a n b n converges.

Problem 2 (WR Ch 3 #9). Find the radius of convergence of each of the following power series:

(a)

n3 zn

Solution. By Theorem 3.39, the radius of convergence of the power series is given by R =

where is defined as
= lim sup
n

p
3
1 3
n
|n 3 | = lim n n = lim n n = 13 = 1,
n

so the radius of convergence is R =

(b)

P 2n

n!

= 1.

zn

Solution. Using the ratio test, the series converges if

a n+1
= lim
lim sup
a n n
n

2n+1
n+1
(n+1)! |z|
2n
n
n! |z|

= lim

2 |z|
< 1,
n +1

which is true for all z C, so the radius of convergence is R = .

(c)

P 2n

n2

zn

Solution. Using Theorem 3.39 again,

s
2n
2
n
= lim sup 2 = lim 2 =
n
n
n
nn

2
lim n

1
n

2 =

2
= 2,
12

so the radius of convergence is R =

(d)

P n3
3n

= 21 .

zn

Solution. By Theorem 3.39,

1 3
s
3
n
lim
n
3
n
13 1
nn
n
n
= lim sup n = lim
=
=
= ,
n 3
3
3
3
3
n
so the radius of convergence is R =

= 3.

Problem 3 (WR Ch 3 #10). Suppose that the coefficients of the power series

a n z n are integers,

infinitely many of which are distinct from zero. Prove that the radius of convergence is at most 1.
Solution. To prove the radius of convergence is at most 1, we must show that if |z| > 1, then

an z n

diverges. Notice that

|z| > 1

|a n z n | = |a n | |z|n > |a n |.

Next, note that if an integer a is nonzero, then |a| 1. Therefore, since there are infinitely many
n N such that a n 6= 0, there are infinitely many n N such that
|a n z n | > |a n | 1,
so lim |a n z n | 6= 0, and thus
n

a n z n diverges by the divergence theorem (3.23).

Problem 4 (WR Ch 3 #11). Suppose a n > 0, s n = a 1 + + a n , and

(a) Prove that

an
1+a n

a n diverges.

diverges.

Solution. Assume (by way of contradiction) that

an
1+a n

converges. Then

an
1+a n

0 by The-

orem 3.23. Since a n 6= 0, we can divide the top and bottom of this fraction by a n to get
1
+1

1
an

0, which implies that

1
an

, which again implies that a n 0. This last result

means that for any > 0, there exists some N1 N such that |a n 0| < for all n N1 . Let
= 1, and choose N1 N so that |a n | < 1 for n N1 .
P an
Next, since we assumed that 1+a
converges, that means the sequence of partial sums
n
Pn
ak
of the series { k=1 1+a } converges in R as we increase n to , so it is a Cauchy sequence.
k

This is equivalent to the statement that for any > 0 there exists some N2 N such that

X
m
n
X
a k
ak

for all n, m N2 ,
<

k=1 1 + a k k=1 1 + a k
or, equivalently,

an
am
++
<
for all n, m N2 .
1 + am
1 + an
Now, if we set N = max(N1 , N2 ), then for any k N we have a k 1, so
an
am
an
am
++
<
++
<
for all n, m N ,
1+1
1 + 1 1 + am
1 + an
P
P
but this proves that a2n is Cauchy, and thus a n converges, contradicting our assumption
that it diverges.
(b) Prove that

and deduce that

a N +1
a N +k
sN
++
1
s N +1
s N +k
s N +k
P an
sn

diverges.

Solution. Since s N +k s N +k1 s N +k2 s N , we have

a N +1
a N +k a N +1
a N +k a N +1 + + a N +k s N +k s N
sN
++

++
=
=
= 1
.
s N +1
s N +k
s N +k
s N +k
s N +k
s N +k
s N +k
P
Now assume (by way of contradiction) that asnn converges. Then its Cauchy, so for any > 0
there is some N N such that
am
an
++
<
sm
sn

for all n, m N .

Letting = 21 , m = N , and n = N + k for some positive integer k, we get

1 a N +1
a N +k
sN
>
++
1
,
2 s N +1
s N +k
s N +k
and doing some arithmetic this means
sN
s N +k

>

1
2

for any positive integer k.

However, since
that

sN
s N +k

a n diverges and a n > 0, we have that s N +k as k , and this implies

0 as k , contradicting the fact that

sN
s N +k

>

1
2

for all k N.

Problem 5 (WR Ch 3 #14). If {s n } is a complex sequence, define its arithmetic means n by

n =

s0 + + sn
n +1

(n = 0, 1, 2, . . .).

(a) If lim s n = s, prove that lim n = s.

Solution. We want to show that for every > 0 there exists an N N such that |n s| <
for all n N . Since we already know that lim s n = s, then there is some N1 N such that
|s n s| <

for all n N1 . Also, since {s n } is a convergent sequence in a metric space (C), its

bounded, so there exists some M > 0 such that |s n s| < M for all n. Putting all this together,
we have

s ++ s

0
n
s
|n s| =
n
+
1

s 0 + + s n (n + 1)s

n +1

(s 0 s) + + (s n s)

n +1
|s 0 s| + + |s n s|

n +1
|s 0 s| + + |s N1 1 s| |s N1 s| + + |s n s|
+
=
n +1
n +1

N1 M (n N1 + 1) 2
<
+
n +1
n +1
N1 M

+ .
n +1 2
In the last step, we let choose N2 to be the smallest positive integer such that N2 >
so that

N1 M
<
n +1

for all n N2 .

Therefore, continuing from before, if we set N = max(N1 , N2 ), then

|n s| <

N1 M
+ < + =
n +1 2 2 2

for all n N ,

finishing the proof that lim n = s.

(b) Construct a sequence {s n } which does not converge, although lim n = 0.
Solution. Let s n = (1)n . Then
s0 + + sn
=
n =
n +1
so lim n = 0, but s n does not converges.
4

1
n+1

if n is even

if n is odd

2N1 M

1,

(c) Can it happen that s n > 0 for all n and that lim sup s n = , although lim n = 0?
Solution. Let
sn =

1
n
1
n

p
+ 3n

if n = 0
if n = k 3 for some integer k
otherwise

p
Since the number of cubic numbers in {1, . . . , n} is given by b 3 nc, which is the largest integer
less than n, then we have
s0 + + sn 1 + n
=
n =
n +1

1
n

p p
+ b 3 nc 3 n

n +1

2+n 3

0
n +1

as n . Therefore lim n = 0. Lastly, we check that

1 p
lim sup s n lim
+ 3 n = .
n n

(d) Put a n = s n s n1 , for n 1. Show that

s n n =

n
1 X
ka k .
n + 1 k=1

()

Assume that lim(na n ) = 0 and that {n } converges. Prove that {s n } converges.

Solution. We prove the first part by induction.
s0 + s1 1
(s 1 s 0 ) = 12 a 1 .
= 2
1 Pn
This establishes the base case. Next, assume that s n n = n+1
ka k , and we have
k=1
s 1 1 = s 1

!
n+1
n
X
X
1
1
ka k =
ka k + (n + 1)(s n+1 s n )
(n + 1) + 1 k=1
n + 2 k=1
1
((n + 1)(s n n ) + (n + 1)(s n+1 s n ))
n +2
1
=
(n + 1)(s n+1 n )
n +2
(n + 1)s n+1 (s 0 + + s n )
=
n +2
(n + 2)s n+1 (s 0 + + s n + s n+1 )
=
n +2
(s 0 + + s n+1 )
= s n+1
n +2

= s n+1 n+1 .
This completes the induction. Now, since {na n } is a complex sequence, and lim na n = 0,
then by part (a) the limit of the arithmetic means of {na n } must also be 0, so to restate this,

!
n
1 X
ka k = lim a n = 0.
lim
n
n n + 1
k=1
5

Therefore,

lim(s n n ) = lim

!
n
1 X
ka k = 0,
n + 1 k=1

by (), and thus lim s n = lim n (and n converges), so finally we have shown that {s n } converges.

Problem 6 (WR Ch 3 #16). Fix a positive number . Choose x 1 >

the recursion formula
x n+1 =

p
, and define x 2 , x 3 , x 4 , . . . by

xn +
.
2
xn

(a) Prove that {x n } decreases monotonically and that lim x n =

p
.

Solution. If a, b R, then
(a b)2 0

a 2 2ab + b 2 0

a2 + b2
ab,
2

with equality holding iff a = b. We will use this last inequality to show that {x n } is bounded
p
below by . We do so by induction. x 0 > is given. Assuming that x n > , we have
p

1
p
p
xn +
x n+1 =
> x n p = .
2
xn
xn
Now we who that {x n } is a decreasing sequence.

1
1
1 x n2
x n x n+1 = x n x n +
xn
=
=
> 0,
2
xn
2
xn
2
xn
using the fact that x n2 > in the last inequality, so we have proven that x n x n+1 > 0 or
equivalently that x n > x n+1 . Since {x n } is a strictly decreasing, bounded sequence, it must
p
converge. Let x = lim x n . Notice that x a > 0, so

1
x n+1 =
xn +
xn +
=
lim x n+1 = lim
n
n 2
2
xn
xn

1
=
x=
x+
2
x
x

=
=
2 2x
x2 =
p
x = .

=
=
p
(b) Put n = x n , and show that
n+1 =
p
so that, setting = 2 ,
n+1 <

2n

2
< pn
2x n 2

2n
(n = 1, 2, 3, . . .).

Solution.
p
p
p
p
x n2 2x n + (x n )2
2
2
1

n+1 = x n+1 = 2 (x n xn ) =
=
= n < pn
2x n
2x n
2x n 2
since x n >

p
. So n+1 <

2n
, and applying this n

n+1 <

2n

<

2n1

times we get

n1
=

2 2

1
< <

2n
.

(c) This is a good algorithm for computing square roots, since the recursion formula is simple
and the convergence is extremely rapid. For example, if = 3 and x 1 = 2, show that 1 / <

1
10

and that therefore

5 = 4 1016 ,

6 < 4 1032 .

Solution.
p
1 2 3
1
1
1
= p = p
.
p =
p <

10
2 3
2 3(2 + 3) 6 + 4 3
24
p
1
< 2 3 1016 < 4 1016 .
5 <

25
p
1
< 2 3 1032 < 4 1032 .
6 <

Problem 7 (WR Ch 3 #18). Replace the recursion formula of Exercise 16 by

x n+1 =

p 1
p+1
xn + xn
p
p

where p is a fixed positive integer, and describe the behavior of the resulting sequences {x n }.
Solution. If the limit exists, let x = lim x n . Then
x n+1 =

p 1
p+1
xn + xn
p
p

lim x n+1 = lim

p 1
p+1
xn + xn
p
p

p 1
x + x p+1
p
p
x

=
p px

x=

=
=

xp =
p
x = p .

=
=

To show the limit exists, well show again the sequence is bounded and decreasing. First we show
its bounded. To do so, well have to use Youngs Inequality. It says that if

1
p

+ q1 = 1 and if a, b > 0,

then

a
q

+ pb a q b p . Thus, letting q =

p
1
p1 , we have p

+ q1 = 1, and so
1

x n+1 =

1
p
p
p 1
p+1
q
xn + xn
> x n p1 = p ,
p
p
p
xn

so {x n } is bounded. Next,
p

x n x n+1 = x n

xn
p 1
p+1 x n

=
> 0,
xn + xn
=
+
p1
p1
p
p
p
px n
px n

so {x n } is decreasing.