1

Solution Manual for

A First Course in Abstract Algebra, with Applications
Third Edition
by Joseph J. Rotman
Exercises for Chapter 1
1.1 True or false with reasons.
(i) There is a largest integer in every nonempty set of negative inte-
gers.
Solution. True. If C is a nonempty set of negative integers, then
C = {n : n C}
is a nonempty set of positive integers. If a is the smallest element
of C, which exists by the Least Integer Axiom, then a c
for all c C, so that a c for all c C.
(ii) There is a sequence of 13 consecutive natural numbers containing
exactly 2 primes.
Solution. True. The integers 48 through 60 form such a sequence;
only 53 and 59 are primes.
(iii) There are at least two primes in any sequence of 7 consecutive
natural numbers.
Solution. False. The integers 48 through 54 are 7 consecutive
natural numbers, and only 53 is prime.
(iv) Of all the sequences of consecutive natural numbers not containing
2 primes, there is a sequence of shortest length.
Solution. True. The set C consisting of the lengths of such (nite)
sequences is a nonempty subset of the natural numbers.
(v) 79 is a prime.
Solution. True.

79 -

81 = 9, and 79 is not divisible by 2, 3,

5, or 7.
(vi) There exists a sequence of statements S(1). S(2). . . . with S(2n)
true for all n 1 and with S(2n 1) false for every n 1.
Solution. True. Dene S(2n 1) to be the statement n = n, and
dene S(2n) to be the statement n = n.
(vii) For all n 0, we have n F
n
, where F
n
is the nth Fibonacci
number.
2
Solution. True. We have 0 = F
0
, 1 = F
1
, 1 = F
2
, and 2 =
F
3
. Use the second form of induction with base steps n = 2 and
n = 3 (verifying the inductive step will show why we choose
these numbers). By the inductive hypothesis, n 2 F
n2
and
n 1 F
n1
. Hence, 2n 3 F
n
. But n 2n 3 for all n 3,
as desired.
(viii) If m and n are natural numbers, then (mn)! = m!n!.
Solution. False. If m = 2 = n, then (mn)! = 24 and m!n! = 4.
1.2 (i) For any n 0 and any r = 1, prove that
1 +r +r
2
+r
3
+ +r
n
= (1 r
n+1
),(1 r).
Solution. We use induction on n 1. When n = 1, both sides
equal 1 +r. For the inductive step, note that
[1 +r +r
2
+r
3
+ +r
n
] +r
n+1
= (1 r
n+1
),(1 r) +r
n+1
=
1 r
n+1
+(1 r)r
n+1
1 r
=
1 r
n+2
1 r
.
(ii) Prove that
1 +2 +2
2
+ +2
n
= 2
n+1
1.
Solution. This is the special case of the geometric series when
r = 2; hence, the sum is (1 2
n+1
),(1 2) = 2
n+1
1. One can
also prove this directly, by induction on n 0.
1.3 Show, for all n 1, that 10
n
leaves remainder 1 after dividing by 9.
Solution. This may be rephrased to say that there is an integer q
n
with
10
n
= 9q
n
+1. If we dene q
1
= 1, then 10 = q
1
+1, and so the base step
is true.
For the inductive step, there is an integer q
n
with
10
n+1
= 10 10
n
= 10(9q
n
+1)
= 90q
n
+10 = 9(10q
n
+1) +1.
Dene q
n+1
= 10q
n
+1, which is an integer.
1.4 Prove that if 0 a b, then a
n
b
n
for all n 0.
Solution. Base step. a
0
= 1 = b
0
, and so a
0
b
0
.
Inductive step. The inductive hypothesis is
a
n
b
n
.
3
Since a is positive, Theorem 1.4(i) gives a
n+1
= aa
n
ab
n
; since b is
positive, Theorem 1.4(i) now gives ab
n
bb
n
= b
n+1
.
1.5 Prove that 1
2
+2
2
+ +n
2
=
1
6
n(n +1)(2n +1) =
1
3
n
3
+
1
2
n
2
+
1
6
n.
Solution. The proof is by induction on n 1. When n = 1, the left side is
1 and the right side is
1
3
+
1
2
+
1
6
= 1.
For the inductive step,
[1
2
+2
2
+ +n
2
] +(n +1)
2
=
1
3
n
3
+
1
2
n
2
+
1
6
n +(n +1)
2
=
1
3
(n +1)
3
+
1
2
(n +1)
2
+
1
6
(n +1).
after some elementary algebraic manipulation.
1.6 Prove that 1
3
+2
3
+ +n
3
=
1
4
n
4
+
1
2
n
3
+
1
4
n
2
.
Solution. Base step: When n = 1, both sides equal 1.
Inductive step:
[1
3
+2
3
+ +n
3
] +(n +1)
3
=
1
4
n
4
+
1
2
n
3
+
1
4
n
2
+(n +1)
3
.
Expanding gives
1
4
n
4
+
3
2
n
3
+
13
4
n
2
+3n +1.
which is
1
4
(n +1)
4
+
1
2
(n +1)
3
+
1
4
(n +1)
2
.
1.7 Prove that 1
4
+2
4
+ +n
4
=
1
5
n
5
+
1
2
n
4
+
1
3
n
3

1
30
n.
Solution. The proof is by induction on n 1. If n 1, then the left side is
1, while the right side is
1
5
+
1
2
+
1
3

1
30
= 1 as well.
For the inductive step,
_
1
4
+2
4
+ +n
4
_
+(n +1)
4
=
1
5
n
5
+
1
2
n
4
+
1
3
n
3

1
30
n +(n +1)
4
.
It is now straightforward to check that this last expression is equal to
1
5
(n +1)
5
+
1
2
(n +1)
4
+
1
3
(n +1)
3

1
30
(n +1).
1.8 Find a formula for 1+3+5+ +(2n1), and use mathematical induction
to prove that your formula is correct.
Solution. We prove by induction on n 1 that the sum is n
2
.
Base Step. When n = 1, we interpret the left side to mean 1. Of course,
1
2
= 1, and so the base step is true.
Inductive Step.
1 +3 +5 + +(2n 1) +(2n +1)
= 1 +3 +5 + +(2n 1)] +(2n +1)
= n
2
+2n +1
= (n +1)
2
.
4
1.9 Find a formula for 1 +

n
j =1
j ! j , and use induction to prove that your
formula is correct.
Solution. A list of the sums for n = 1. 2. 3. 4. 5 is 2. 6. 24. 120. 720.
These are factorials; better, they are 2!. 3!. 4!. 5!. 6!. We have been led to
the guess
S(n) : 1 +
n

j =1
j ! j = (n +1)!.
We now use induction to prove that the guess is always true. The base step
S(1) has already been checked; it is on the list. For the inductive step, we
must prove
S(n +1) : 1 +
n+1

j =1
j ! j = (n +2)!.
Rewrite the left side as
_
1 +
n

j =1
j ! j
_
+(n +1)!(n +1).
By the inductive hypothesis, the bracketed term is (n +1)!, and so the left
side equals
(n +1)! +(n +1)!(n +1) = (n +1)![1 +(n +1)]
= (n +1)!(n +2)
= (n +2)!.
By induction, S(n) is true for all n 1.
1.10 (M. Barr) There is a famous anecdote describing a hospital visit of G. H.
Hardy to Ramanujan. Hardy mentioned that the number 1729 of the taxi
he had taken to the hospital was not an interesting number. Ramanujan
disagreed, saying that it is the smallest positive integer that can be written
as the sum of two cubes in two different ways.
(i) Prove that Ramanujans statement is true.
Solution. First, 1729 is the sum of two cubes in two different
ways:
1729 = 1
3
+12
3
; 1927 = 9
3
+10
3
.
Second, no smaller number n has this property. If n = a
3
+ b
3
,
then a. b 12. It is now a matter of checking all pairs a
3
+b
3
for
such a and b.
5
(ii) Prove that Ramanujans statement is false.
Solution. One must pay attention to hypotheses. Consider a
3
+b
3
if b is negative:
728 = 12
3
+(10
3
) = 9
3
+(1)
3
.
1.11 Derive the formula for

n
i =1
i by computing the area (n +1)
2
of a square
with sides of length n +1 using Figure 1.1.
Solution. Compute the area A of the square in two ways. On the one hand,
A = (n +1)
2
. On the other hand, A = |D| +2|S|, where D is the diagonal
and S is the staircase. Therefore,
|S| =
1
2
_
(n +1)
2
(n +1)
_
=
1
2
n(n +1).
But |S| is the sum we are seeking.
1
2
3
4
5 1 1 1 1 1
1 1 1 1
1 1
1 1
1
1
Figure 1.1
1 +2 + +n =
1
2
(n
2
+n)
1 1 1 1 1
1 1 1 1
1 1
1 1
1
1
Figure 1.2
1 +2 + +n =
1
2
n(n +1)
1.12 (i) Derive the formula for

n
i =1
i by computing the area n(n +1) of
a rectangle with height n +1 and base n, as pictured in Figure 1.2.
Solution. Compute the area R of the rectangle in two ways. On
the one hand, R = n(n +1). On the other hand, R = 2|S|, where
S is the shaded region (whose area is what we seek).
(ii) (Alhazen) For xed k 1, use Figure 1.3 to prove
(n +1)
n

i =1
i
k
=
n

i =1
i
k+1
+
n

i =1
_
i

p=1
p
k
_
.
6
Solution. As indicated in Figure 1.3, a rectangle with height n +1
and base

n
i =1
i
k
can be subdivided so that the shaded staircase
has area

n
i =1
i
k+1
, while the area above it is
1
k
+(1
k
+2
k
) +(1
k
+2
k
+3
k
) + +(1
k
+2
k
+ +n
k
).
One can prove this, for xed k, by induction on n 1.
1 2 3 5 4
k k k k k
1
5 4 +
k k
k
1
2
3
5
4
k+1
k+1
k+1
k+1
k+1
1 + 2
3
4
k
k
k
k
3
k
1 + 2
k k
1 + 2 + +
k k
1 + 2 + +
k k
k
3 +
Figure 1.3 Alhazans Dissection
(iii) Given the formula

n
i =1
i =
1
2
n(n +1), use part (ii) to derive the
formula for

n
i =1
i
2
.
Solution.
(n +1)
n

i =0
i =
n

i =0
i
2
+
n

i =0
_
i

p=0
p
_
=
n

i =0
i
2
+
n

i =0
1
2
i (i +1)
=
n

i =0
i
2
+
1
2
n

i =0
i
2
+
1
2
n

i =0
i.
Therefore,
(n +1
1
2
)
n

i =0
i =
3
2
n

i =0
i
2
.
and so
n

i =0
i
2
=
2
3
(n +
1
2
)
1
2
n(n +1)
=
1
3
1
2
(2n +1)n(n +1)
=
1
6
(2n +1)n(n +1).
7
1.13 (i) Prove that 2
n
> n
3
for all n 10.
Solution. Base step. 2
10
= 1024 > 10
3
= 1000. (Note that
2
9
= 512 - 9
3
= 729.)
Inductive step Note that n 10 implies n 4. The inductive
hypothesis is 2
n
> n
3
; multiplying both sides by 2 gives
2
n+1
= 2 2
n
> 2n
3
= n
3
+n
3
n
3
+4n
2
= n
3
+3n
2
+n
2
> n
3
+3n
2
+4n
= n
3
+3n
2
+3n +n
n
3
+3n
2
+3n +1
= (n +1)
3
.
(ii) Prove that 2
n
> n
4
for all n 17.
Solution. Base step. 2
17
= 131. 072 > 17
4
= 83. 521. (Note that
16
4
= (2
4
)
4
= 2
16
.)
Inductive step. Note that n 17 implies n 7. The inductive
hypothesis is 2
n
> n
4
; multiplying both sides by 2 gives
2
n+1
= 2 2
n
> 2n
4
= n
4
+n
4
n
4
+5n
3
n
4
+4n
3
+n
3
n
4
+4n
3
+7n
2
n
4
+4n
3
+6n
2
+n
2
n
4
+4n
3
+6n
2
+5n
n
4
+4n
3
+6n
2
+4n +1 = (n +1)
4
.
1.14 Around 1350, N. Oresme was able to sum the series

n=1
n,2
n
by dis-
secting the region in Figure 1.4 in two ways. Let A
n
be the vertical rect-
angle with base
1
2
n
and height n, so that area(A
n
) = n,2
n
, and let B
n
be
horizontal rectangle with base
1
2
n
+
1
2
n+1
+ and height 1. Prove that

n=1
n,2
n
= 2.
Solution. You may assume that

n=0
ar
n
= a,(1r) if 0 r - 1. Now
compute the area using

A
n
=

B
n
.
8
A
1
A
3
A
2
1
2
1
4
1
8
1
B
0
B
1
B
2
B
3
1
1
1
1
Figure 1.4 Oresmes Dissections
1.15 Let g
1
(x). . . . . g
n
(x) be differentiable functions, and let f (x) be their prod-
uct: f (x) = g
1
(x) g
n
(x). Prove, for all integers n 2, that the deriva-
tive
f

(x) =
n

i =1
g
1
(x) g
i 1
(x)g

i
(x)g
i +1
(x) g
n
(x).
Solution. Base step. If n = 2, this is the usual product rule for derivatives.
Inductive step. Dene h(x) = g
1
(x) g
n
(x) = f (x),g
n+1
(x). Rewrite
what has to be shown:
f

(x) =
n+1

j =1
g

j
(x) f (x)
g
j
(x)
.
Now
f

(x) = (h(x)g
n+1
(x))

= h

(x)g
n+1
(x) +h(x)g

n+1
(x)
=
n

= 1
i =1
_
g

i
(x)h(x)
g
i
(x)
_
g
n+1
(x) +
_
f (x)
g
n+1
(x)
_
g

n
=
n+1

j =1
g

j
(x) f (x)
g
j
(x)
.
1.16 Prove, for every n N, that (1 + x)
n
1 + nx whenever x R and
1 + x > 0.
Solution. We prove the inequality by induction on n 1. The base step
n = 1 says 1 + x 1 + x, which is obviously true. For the inductive step,
9
we record the inductive hypothesis:
(1 + x)
n
1 +nx.
Multiplying both sides of this inequality by the positive number 1 + x
preserves the inequality:
(1 + x)
n+1
= (1 + x)(1 + x)
n
(1 + x)(1 +nx)
= 1 +(n +1)x +nx
2
1 +(n +1)x.
because nx
2
0.
1.17 Prove that every positive integer a has a unique factorization a = 3
k
m,
where k 0 and m is not a multiple of 3.
Solution. Model your solution on the proof of Proposition 1.14. Replace
even by multiple of 3 and odd by not a multiple of 3. We prove
this by the second form of induction on a 1. The base step n = 1 holds,
for 1 = 3
0
1 is a factorization of the desired kind.
For the inductive step, let a 1. If a is not a multiple of 3, then a = 3
0
a
is a good factorization. If a = 3b, then b - a, and so the inductive
hypothesis gives k 0 and an integer c not divisible by 3 such that b =
3
k
c. It follows that a = 3b = 3
k+1
c, which is a factorization of the desired
kind. We have proved the existence of a factorization.
To prove uniqueness, suppose that n = 3
k
m = 3
t
m

, where both k and

t are nonnegative and both m and m

are not multiples of 3; it must be

shown that k = t and m = m

. We may assume that k t . If k > t , then

canceling 3
t
from both sides gives 3
kt
m = m

. Since k t > 0, the left

side is a multiple of 3 while the right side is not; this contradiction shows
that k = t . We may thus cancel 3
k
from both sides, leaving m = m

.
1.18 Prove that F
n
- 2
n
for all n 0, where F
0
. F
1
. F
2
. . . . is the Fibonacci
sequence.
Solution. The proof is by the second form of induction.
Base step:
F
0
= 0 - 1 = 2
0
and F
1
= 1 - 2 = 2
1
. (There are two base steps
because we will have to use two predecessors for the inductive step.)
Inductive step:
10
If n 2, then
F
n
= F
n1
+ F
n2
- 2
n1
+2
n2
(by inductive hypothesis)
- 2
n1
+2
n1
= 2 2
n1
= 2
n
.
By induction, F
n
- 2
n
for all n 0.
Notice that the second form is the appropriate induction here, for we are
using two predecessors, S(n 2) and S(n 1), to prove S(n).
1.19 If F
n
denotes the nth term of the Fibonacci sequence, prove that
m

n=1
F
n
= F
m+2
1.
Solution. By Theorem 1.15, we have F
n
=
1

5
(
n

n
) for all n. Hence,
m

n=1
F
n
=
m

n=1
1

5
(
n

n
)
=
1

5
m

n=1
(
n

n
)
=
1

5
__
1
m+1
1
1
_

_
1
m+1
1
1
__
.
Now ( 1) = 1, so that 1,(1 ) = ; similarly, 1,(1 ) = .
Therefore,
m

n=1
F
n
=
1

5
__
1
m+1
1
1
_

_
1
m+1
1
1
__
=
1

5
_
()[(1
m+1
) 1] [()(1
m+1
) 1]
_
=
1

5
[( ) +(
m+2

m+2
)].
This is the desired formula, for
1

5
( ) = 1 and
1

5
(
m+2

m+2
) =
F
m+2
.
1.20 Prove that 4
n+1
+5
2n1
is divisible by 21 for all n 1.
Solution. We use the second form of induction.
11
Base Step. If n = 1, then
4
n+1
+5
2n1
= 16 +5 = 21.
which is obviously divisible by 21. Since our inductive step will involve
two predecessors, we are obliged to check the case n = 2. But 4
3
+ 5
3
=
64 +125 = 189 = 21 9.
Inductive Step.
4
n+2
+5
2n+1
= 4 4
n+1
+5
2
5
2n1
= 4 4
n+1
+(4 5
2n1
4 5
2n1
) +5
2
5
2n1
= 4(4
n+1
+5
2n1
) +5
2n1
(5
2
4).
Now the last term is divisible by 21; the rst term, by the inductive hypoth-
esis, and the second because 5
2
4 = 21.
1.21 For any integer n 2, prove that there are n consecutive composite num-
bers. Conclude that the gap between consecutive primes can be arbitrarily
large.
Solution. The proof has nothing to do with induction. If 2 a n + 1,
then a is a divisor of (n + 1)!; say, (n + 1)! = da for some integer d. It
follows that (n +1)! +a = (d +1)a, and so (n +1)! +a is composite for
all a between 2 and n +1.
1.22 Prove that the rst and second forms of mathematical induction are equiv-
alent; that is, prove that Theorem 1.4 is true if and only if Theorem 1.12 is
true.
Solution. Absent.
1.23 (Double Induction) Let S(m. n) be a doubly indexed family of statements,
one for each m 0 and n 0. Suppose that
(i) S(0. 0) is true;
(ii) if S(m. 0) is true, then S(m +1. 0) is true;
(iii) if S(m. n) is true for all m 0, then S(m. n +1) is true for all m 0.
Prove that S(m. n) is true for all m 0 and n 0.
Solution. Conditions (i) and (ii) are the hypotheses needed to prove, by
(ordinary) induction that the statements S(m. 0) are true for all m 0.
Now consider the statements
T(n) : S(m. n) is true for all m 0.
We prove that all the statements T(n) are true by induction on n 0. The
base step has been proved above, and condition (iii) is precisely what is
needed for the inductive step.
12
1.24 Use double induction to prove that
(m +1)
n
> mn
for all m. n 0.
Solution. According to Exercise 1.23, there are three things to verify.
(i) S(0. 0): (0 +0)
0
= 1 0.
(ii) S(m. 0) S(m+1. 0): if (m+1)
0
> m, then (m+2)
0
> (m+1)0 =
0?
(iii) S(m. n) S(m. n + 1): does (m + 1)
n
> mn imply (m + 1)
n+1
>
m(n +1)? Yes, because
(m +1)
n+1
= (m +1)(m +1)
n
> (m +1)mn
= m
2
n +mn
> mn +m.
for m
2
n mn and mn m.
Notice that 2
n
> n is the special case S(0. n).
1.25 For every acute angle , i.e., 0

- - 90

, prove that
sin +cot +sec 3.
Solution. That is an acute angle implies that the numbers sin , cot ,
and sec are all positive. The inequality of the means gives
_
1
3
(sin +cot +sec )
_
3
sin cot sec .
Now
sin cot sec = sin
cos
sin
1
cos
= 1.
so that
_
1
3
(sin +cot +sec )
_
3
1 and
1
3
(sin +cot +sec ) 1.
Therefore, sin +cot +sec 3.
1.26 Isoperimetric Inequality.
(i) Let p be a positive number. If L is an equilateral triangle with
perimeter p = 2s, prove that area(L) = s
2
,

27.
Solution. This is an elementary fact of high school geometry.
13
(ii) Of all the triangles in the plane having perimeter p, prove that the
equilateral triangle has the largest area.
Solution. Use Herons formula: if a triangle T has area A and
sides of lengths a. b. c, then
A
2
= s(s a)(s b)(s c).
where s =
1
2
(a +b +c). The inequality of the means gives
_
(s a) +(s b) +(s c)
3
_
3
(s a)(s b)(s c) =
A
2
s
.
with equality holding if and only if s a = s b = s c. Thus,
equality holds if and only if a = b = c, which is to say T is
equilateral.
1.27 Prove that if a
1
. a
2
. . . . . a
n
are positive numbers, then
(a
1
+a
2
+ +a
n
)(1,a
1
+1,a
2
+ +1,a
n
) n
2
.
Solution. By the inequality of the means, [(a
1
+ a
2
+ + a
n
),n]
n

a
1
a
n
and [(1,a
1
+1,a
2
+ +1,a
n
),n]
n
1,a
1
1,a
n
. Now use
the general fact that if p q > 0 and p

> 0, then pp

qq

to obtain (a
1
+ a
2
+ + a
n
),n]
n
[(1,a
1
+ 1,a
2
+ + 1,a
n
),n]
n

(a
1
a
n
)(1,a
1
1,a
n
). But the right side is a
1
a
n
(1,a
1
) (1,a
n
) =
1, so that
(a
1
+a
2
+ +a
n
),n]
n
[(1,a
1
+1,a
2
+ +1,a
n
),n]
n
1.
Taking nth roots gives (a
1
+a
2
+ +a
n
)(1,a
1
+1,a
2
+ +1,a
n
) n
2
.
1.28 True or false with reasons.
(i) For all integers r with 0 - r - 7, the binomial coefcient
_
7
r
_
is a
multiple of 7.
Solution. True.
(ii) For any positive integer n and any r with 0 - r - n, the binomial
coefcient
_
n
r
_
is a multiple of n.
Solution. False.
(iii) Let D be a collection of 10 different dogs, and let C be a collection
of 10 different cats. There are the same number of quartets of dogs
as there are sextets of cats.
Solution. True.
(iv) If q is a rational number, then e
2i q
is a root of unity.
Solution. True.
14
(v) Let f (x) = ax
2
+ bx + c, where a. b. c are real numbers. If z is
a root of f (x), then z is also a root of f (x).
Solution. True.
(vi) Let f (x) = ax
2
+bx +c, where a. b. c are complex numbers. If
z is a root of f (x), then z is also a root of f (x).
Solution. False.
(vii) The primitive 4th roots of unity are i and i .
Solution. True.
1.29 Prove that the binomial theorem holds for complex numbers: if u and : are
complex numbers, then
(u +:)
n
=
n

r=0
_
n
r
_
u
nr
:
r
.
Solution. The proof of the binomial theorem for real numbers used only
properties of the three operations: addition, multiplication, and division.
These operations on complex numbers have exactly the same properties.
(Division enters in only because we chose to expand (a +b)
n
by using the
formula for (1 +x)
n
; had we not chosen this expository path, then division
would not have been used. Thus, the binomial theorem really holds for
commutative rings.)
1.30 Show that the binomial coefcients are symmetric:
_
n
r
_
=
_
n
n r
_
for all r with 0 r n.
Solution. By Lemma 1.17, both
_
n
r
_
and
_
n
nr
_
are equal to
n!
r!(n r)!
.
1.31 Show, for every n, that the sum of the binomial coefcients is 2
n
:
_
n
0
_
+
_
n
1
_
+
_
n
2
_
+ +
_
n
n
_
= 2
n
.
Solution. By Corollary 1.19, if f (x) = (1 + x)
n
, then there is the expan-
sion
f (x) =
_
n
0
_
+
_
n
1
_
x +
_
n
2
_
x
2
+ +
_
n
n
_
x
n
.
Evaluating at x = 1 gives the answer, for f (1) = (1 +1)
n
= 2
n
.
15
1.32 (i) Show, for every n 1, that the alternating sum of the binomial
coefcients is zero:
_
n
0
_

_
n
1
_
+
_
n
2
_
+(1)
n
_
n
n
_
= 0.
Solution. If f (x) = (1 + x)
n
, then f (1) = (1 1)
n
= 0; but
the expansion is the alternating sum of the binomial coefcients.
(ii) Use part (i) to prove, for a given n, that the sum of all the binomial
coefcients
_
n
r
_
with r even is equal to the sum of all those
_
n
r
_
with r odd.
Solution. By part (i),
_
n
0
_

_
n
1
_
+
_
n
2
_

_
n
n
_
= 0.
Since the sign of
_
n
r
_
is (1)
r
, the terms
_
n
r
_
with r even are posi-
tive while those with r odd are negative. Just put those coefcients
with negative coefcient on the other side of the equation.
1.33 Prove that if n 2, then
n

r=1
(1)
r1
r
_
n
r
_
= 0.
Solution. Again, consider f (x) = (1+x)
n
. There are two ways to describe
its derivative f

(x). On the one hand, f

(x) = n(1 + x)
n1
. On the other
hand, we can do term-by-term differentiation:
f

(x) =
n

r=1
r
_
n
r
_
x
r1
.
Evaluating at x = 1 gives f

(1) = n(1 1)
n1
= 0, since n 1 1.
On the other hand, we can use the expansion to see
f

(1) =
n

r=1
(1)
r1
r
_
n
r
_
.
1.34 If 1 r n, prove that
_
n
r
_
=
n
r
_
n 1
r 1
_
.
Solution. Absent.
16
1.35 Let
1
. . . . .
n
be complex numbers with |
j
| = 1 for all j , where n 2.
(i) Prove that

j =1

j =1

= n.
Solution. The triangle inequality gives |u +:| |u| +|:| for all
complex numbers u and :, with no restriction on their norms. The
inductive proof is routine.
(ii) Prove that there is equality,

j =1

= n.
if and only if all the
j
are equal.
Solution. The proof is by induction on n 2.
For the base step, suppose that |
1
+
2
| = 2. Therefore,
4 = |
1
+
2
|
2
= (
1
+
2
) (
1
+
2
)
= |
1
|
2
+2
1

2
+|
2
|
2
= 2 +2
1

2
.
Therefore, 2 = 1 +
1

2
, so that
1 =
1

2
= |
1
||
2
| cos
= cos .
where is the angle between
1
and
2
(for |
1
| = 1 = |
2
|).
Therefore, = 0 or = , so that
2
=
1
. We cannot have

2
=
1
, for this gives |
1
+
2
| = 0.
For the inductive step, let |

n+1
j =1

j
| = n +1. If |

n
j =1

j
| - n,
then part (i) gives

_
n

j =1

j
_
+
n+1

j =1

+1 - n +1.
contrary to hypothesis. Therefore, |

n
j =1

j
| = n, and so the
inductive hypothesis gives
1
. . . . .
n
all equal, say to . Hence,

n
j =1

j
= n, and so
|n +
n+1
| = n +1.
17
The argument concludes as that of the base step.
(n +1)
2
= (n +
n+1
) (n +
n+1
)
= n
2
+2n
n+1
+1.
so that
n+1
= 1. By the base step, =
n+1
, and the proof is
complete.
1.36 (Star of David) Prove, for all n > r 1, that
_
n 1
r 1
__
n
r +1
__
n +1
r
_
=
_
n 1
r
__
n
r 1
__
n +1
r +1
_
.
_
n 1
r 1
_
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
_
n 1
r
_
_
n
r 1
_
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
U
_
n
r
_ _
n
r +1
_
_
n +1
r
_
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
_
n +1
r +1
_
Solution. Using Pascals formula, one sees that both sides are equal to
(n 1)!n!(n +1)!
(r 1)!r!(r +1)!(n r 1)!(n r)!(n r +1)!
.
1.37 For all odd n 1, prove that there is a polynomial g
n
(x), all of whose
coefcients are integers, such that
sin(nx) = g
n
(sin x).
Solution. From De Moivres theorem,
cos nx +i sin nx = (cos x +i sin x)
n
.
we have
sin nx = Im (cos x +i sin x)
n
= Im
_
n

r=0
_
n
r
_
i
r
sin
r
x cos
nr
x
_
.
18
Write n = 2m + 1. Only odd powers of i are imaginary, so that, if r =
2k +1,
sin nx =

0km
_
n
2k +1
_
(1)
k
sin
2k+1
x cos
2(mk)
x.
But
cos
2(mk)
x = (cos
2
x)
mk
= (1 sin
2
x)
mk
.
and so we have expressed sin nx as a polynomial in sin x.
1.38 (i) What is the coefcient of x
16
in (1 + x)
20
?
Solution. Pascals formula gives
_
20
16
_
= 4845.
(ii) How many ways are there to choose 4 colors from a palette con-
taining paints of 20 different colors?
Solution. Pascals formula gives
_
20
4
_
= 4845. One could also
have used part (i) and Exercise 1.30.
1.39 Give at least two different proofs that a set X with n elements has exactly
2
n
subsets.
Solution. There are many proofs of this. We offer only three.
Algebraic.
Let X = {a
1
. a
2
. . . . . a
n
}. We may describe each subset S of X by a
bitstring; that is, by an n-tuple
(c
1
. c
2
. . . . . c
n
).
where
c
i
=
_
0 if a
i
is not in S
1 if a
i
is in S.
(after all, a set is determined by the elements comprising it). But there are
exactly 2
n
such n-tuples, for there are two choices for each coordinate.
Combinatorial.
Induction on n 1 (taking base step n = 0 is also ne; the only set with
0 elements is X = , which has exactly one subset, itself). If X has just
one element, then there are two subsets: and X. For the inductive step,
assume that X has n + 1 elements, of which one is colored red, the other
n being blue. There are two types of subsets S: those that are solid blue;
those that contain the red. By induction, there are 2
n
solid blue subsets;
denote them by B. But, there are as many subsets R containing the red
as there are solid blue subsets: each R arises by adjoining the red element
to a solid blue subset, namely, B = R {red} (even the singleton subset
consisting of the red element alone arises in this way, by adjoining the red
element to ). Hence, there are 2
n
+2
n
= 2
n+1
subsets.
19
Binomial Coefcients.
If X has n elements, then the number of its subsets is the sum of the
number of 0-subsets (there is only 1, the empty set), the number of 1-
subsets, the number of 2-subsets, etc. But
_
n
r
_
is the number of r-subsets,
as we have seen in the text, and so the total number of subsets is the sum
of all the binomial coefcients, which is 2
n
, by Exercise 1.31.
1.40 A weekly lottery asks you to select 5 different numbers between 1 and 45.
At the weeks end, 5 such numbers are drawn at random, and you win the
jackpot if all your numbers match the drawn numbers. What is your chance
of winning?
Solution. The answer is 45 choose 5, which is
_
45
5
_
= 1. 221. 759. The
odds against your winning are more than a million to one.
1.41 Assume that term-by-term differentiation holds for power series: if f (x)
= c
0
+c
1
x +c
2
x
2
+ +c
n
x
n
+ , then the power series for the derivative
f

(x) is
f

(x) = c
1
+2c
2
x +3c
3
x
2
+ +nc
n
x
n1
+ .
(i) Prove that f (0) = c
0
.
Solution. f (0) = c
0
, for all the other terms are 0. (If one
wants to be fussythis is the wrong course for analytic fussiness
then the partial sums of the series form the constant sequence
c
0
. c
0
. c
0
. . . ..)
(ii) Prove, for all n 0, that
f
(n)
(x) = n!c
n
+(n +1)!c
n+1
x + x
2
g
n
(x).
where g
n
(x) is some power series .
Solution. This is a straightforward proof by induction on n 0.
The base step is obvious; for the inductive step, just observe that
f
n+1
(x) = ( f
(n)
(x))

. As f
(n)
(x) is a power series, by assump-
tion, its derivative is computed term by term.
(iii) Prove that c
n
= f
(n)
(x)(0),n! for all n 0. (Of course, this is
Taylors formula.)
Solution. If n = 0, then our conventions that f
(0)
(x) = f (x) and
0! = 1 give the result. For the inductive step, use parts (i) and (ii).
1.42 (Leibniz) Prove that if f and g are C

-functions, then
( f g)
(n)
(x) =
n

k=0
_
n
k
_
f
(k)
(x) g
(nk)
(x).
Solution. The proof is by induction on n 1.
20
If n = 1, then the equation is precisely the product rule of Calculus:
( f g)

= f

g + f g

. For the inductive step, we have

( f g)
n+1
= [( f g)
n
]

=
_
n

k=0
_
n
k
_
f
k
g
nk
_

=
n

k=0
_
n
k
_ _
f
k
g
nk
_

=
n

k=0
_
n
k
_ _
f
k+1
g
nk
+ f
k
g
nk+1
_
=
n

k=0
_
n
k
_
f
k+1
g
nk
+
n

k=0
_
n
k
_
f
k
g
nk+1
.
Rewrite this last expression without the sigma notation:
_
n
0
_
f
0
g
n+1
+
_
n
1
_
f
1
g
n
+ +
_
n
k
_
f
k
g
nk+1
+
+
_
n
0
_
f
1
g
n
+ +
_
n
k 1
_
f
k
g
nk+1
.
The coefcient of f
k
g
nk+1
is thus
_
n
k1
_
+
_
n
k
_
=
_
n+1
k
_
, by Lemma 1.17,
as desired.
1.43 Find

i .
Solution. By De Moivres theorem, since i = e
i ,2
, we have

i = e
i ,4
= cos ,4 +i sin ,4 =

2
2
+i

2
2
.
1.44 (i) If z = r[cos +i sin ], show that
n =
n

r [cos(,n) +i sin(,n)]
is an nth root of z, where r 0.
Solution. By De Moivres theorem,
n
n
= (
n

r)
n
([cos(,n) +i sin(,n)])
n
= r[cos() +i sin()].
(ii) Show that every nth root of z has the form
k
n, where is a
primitive nth root of unity and k = 0. 1. 2. . . . . n 1.
21
Solution. If a
n
= z = b
n
, then 1 = a
n
,b
n
= (a,b)
n
. Therefore,
a,b is an nth root of unity, and so a,b =
k
for some k; that is,
a =
k
b. In particular, if b = n, then b =
k
n.
1.45 (i) Find

8 +15i .
Solution. The polar coordinates of (8. 15) are (17. 62

), and sin 31

.515 and cos 31

.857. Hence, 8+15i 17(cos 62

+i sin 62

),
and so

8 +15i 17(cos 31

+i sin 31

) 3.533+2.123i . (Of course,

the other square root is the negative of this one.)
(ii) Find all the fourth roots of 8 +15i .
Solution. sin 15.5

.267 and cos 15.5

.967. Hence,
4

8 +15i =
4

17(cos 15.5

+i sin 15.5

) 1.964 +542i.
By Exercise 1.44, the other fourth roots are obtained by multiply-
ing this one by i , 1, and i .
1.46 True or false with reasons.
(i) 6 | 2.
Solution. False.
(ii) 2 | 6.
Solution. True.
(iii) 6 | 0.
Solution. True.
(iv) 0 | 6.
Solution. False.
(v) 0 | 0.
Solution. True.
(vi) (n. n +1) = 1 for every natural number n.
Solution. True.
(vii) (n. n +2) = 2 for every natural number n.
Solution. False.
(viii) If b and m are positive integers, then b | m if and only if the last
b-adic digit d
0
of m is 0.
Solution. True.
(ix) 113 is a sum of distinct powers of 2.
Solution. True.
22
(x) If a and b are natural numbers, there there are natural numbers s
and t with gcd(a. b) = sa +t b.
Solution. False.
1.47 Given integers a and b (possibly negative) with a = 0, prove that there
exist unique integers q and r with b = qa +r and 0 r - |a|.
Solution. We have already proved this when a > 0 and b 0. Assume
now that a > 0 and b - 0. Now b > 0, and so there are integers q and r
with b = qa +r and 0 r - a; it follows that b = qb r. If r = 0,
we are done; if r > 0, then b = (q 1)a +(a r) and 0 - a r - a
(by Proposition A.2(ii), r - 0 implies a r - a.
Now assume that a - 0, so that a > 0 (and so |a| = a. By what we
have proved so far, there are integers q and r with b = q(a) +r, where
0 r - a; that is, b = (q)a +r, where 0 r - |a|.
1.48 Prove that

2 is irrational using Proposition 1.14 instead of Euclids lemma.

Solution. Assume, on the contrary, that

2 = a,b, where a and b are inte-

gers. By Proposition 1.14, we have a = 2
k
m and b = 2

n, where k. 0
and m. n are odd. If k , then we may cancel to obtain

2 = 2
k
m,n;
otherwise,

2 = m,2
k
m. We may assume, therefore, that

2 = a,b
where at least one of a or b is odd. Squaring and cross-multiplying, we
have 2b
2
= a
2
. Hence, a
2
is even; it follows that a itself is even, for odd
odd is odd. Write a = 2c, so that 2b
2
= a
2
= 4c
2
. Thus, b
2
= 2c
2
, which
implies, as above, that b is even. This contradicts the fact that at least one
of a or b is odd.
1.49 Let p
1
. p
2
. p
3
. . . . be the list of the primes in ascending order: p
1
= 2,
p
2
= 3, p
3
= 5, and so forth. Dene f
k
= p
1
p
2
p
k
+ 1 for k 1.
Find the smallest k for which f
k
is not a prime.
Solution. f
1
. f
2
. f
3
. f
4
, and f
5
are prime, but
f
6
= 2 3 5 7 11 13 +1 = 30031 = 59 509.
1.50 Prove that if d and d

are nonzero integers, each of which divides the other,

then d

= d.
Solution. Assume that d = ad

and d

= bd. Then
d = ad

= abd.
so that canceling d gives 1 = ab. As a and b are nonzero integers, |a| 1
and |b| 1. But 1 = |ab| = |a| |b| gives |a| = 1 = |b|. Hence, a = 1 = b
or a = 1 = b.
1.51 If is a root of unity, prove that there is a positive integer d with
d
= 1
such that whenever
k
= 1, then d | k.
23
Solution. Dene I = {k in Z :
k
= 1}. Of course, I is a subset of Z, and
it contains positive numbers because is a root of unity. Observe that
(i) If k is in I and u is in Z, then uk is in I :
If k is in I , then
k
= 1. Hence,
uk
= (
k
)
u
= 1
u
= 1, and so
uk is in I .
(ii) If k. are in I , then k + is in I :
If k. are in I , then
k
= 1 =

, so that 1 =
k

=
k+
;
hence, k + is in I .
It now follows from Corollary 1.37 that there is a positive number d in I ,
i.e.,
d
= 1, such that every k in I is a multiple of d; that is, if
k
= 1,
then d | k.
1.52 Show that every positive integer m can be written as a sum of distinct pow-
ers of 2; show, moreover, that there is only one way in which m can so be
written.
Solution. In base 2, the only digits are 0 and 1. If we neglect the binary
digits 0, then every positive integer is uniquely a sum of powers of 2.
1.53 Find the b-adic digits of 1000 for b = 2, 3, 4, 5, and 20.
Solution.
base 2 : 1000 = 1111101000
base 3 : 1000 = 1101001
base 4 : 1000 = 33220
base 5 : 1000 = 13000
base 20 : 1000 = 2.
where is a symbol denoting the new digit ten.
1.54 (i) Prove that if n is squarefree (i.e., n > 1 and n is not divisible by
the square of any prime), then

n is irrational.
Solution. We rewrite the proof of Proposition 1.14. Suppose, on
the contrary, that

n is rational, where n is squarefree; that is,

n = a,b. We may assume that a,b is in lowest terms; that is,

(a. b) = 1. Squaring, a
2
= nb
2
. Let p be a prime divisor of n,
so that n = pq. Since n is squarefree, ( p. q) = 1. By Euclids
lemma, p | a, so that a = pm, hence p
2
m
2
= a
2
= pqb
2
, and
pm
2
= qb
2
. By Euclids lemma, p | b, contradicting (a. b) = 1.
(ii) Prove that
3

2 is irrational.
Solution. Assume that
3

4 = a,b, where (a. b) = 1. Then 4b

3
=
a
3
, so that a is even; say, a = 2m. Hence 4b
3
= 8m
3
; canceling,
b
3
= 2m
3
, forcing b to be even. This contradicts (a. b) = 1.
24
1.55 (i) Find d = gcd(12327. 2409), nd integers s and t with d = 12327s+
2409t , and put the fraction 2409/12327 in lowest terms.
Solution. One uses the Euclidean algorithmto get: (12327. 2409) =
3 and 3 = 12327 299 2409 1530; the fraction 2409/12327 =
803/4109 is in lowest terms.
(ii) Find d = gcd(7563. 526), and express d as a linear combination
of 7563 and 526.
Solution. The Euclidean algorithm gives
(7563. 526) = 1 and 1 = 532 526 37 7563.
(iii) Find d = gcd(73122. 7404621) and express d as a linear combi-
nation of 73122 and 7404621.
Solution. Here are the equations of the Euclidean algorithm:
7404621 = 101 73122 +19299
73122 = 3 19299 +15225
19299 = 1 15225 +4074
15225 = 3 4074 +3003
4074 = 1 3003 +1071
3003 = 2 1071 +861
1071 = 1 861 +210
861 = 4 210 +21
210 = 10 21.
We conclude that the gcd is 21. Following the algorithm in the
text, we nd that
21 = 34531 73122 341 7404621.
1.56 Let a and b be integers, and let sa +t b = 1 for s, t in Z. Prove that a and
b are relatively prime.
Solution. If sa +t b = 1, then any common divisor of a and b must divide
1; hence, a and b are relatively prime.
1.57 If d = (a. b), prove that a,d and b,d are relatively prime.
Solution. Absent.
1.58 Prove that if (r. m) = 1 = (r

. m), then (rr

. m) = 1.
Solution. Since (r. m) = 1, we have ar + bm = 1; since (r

. m) = 1, we
have sr

+t m = 1. Multiplying,
1 = (ar +bm)(sr

+t m) = (as)rr

+(art +bsr

+bt m)m.
25
Therefore, 1 is a linear combination of rr

and m; as 1 is obviously the

smallest positive linear combination, it must be their gcd.
1.59 Let a. b and d be integers. If d = sa +t b, where s and t are integers, nd
innitely many pairs of integers (s
k
. t
k
) with d = s
k
a +t
k
b.
Solution. If d = sa + t b, and if we dene s
k
= s + kb and t
k
= t ka,
then d = s
k
a +t
k
b for all k.
1.60 If a and b are relatively prime and if each divides an integer n, prove that
their product ab also divides n.
Solution. Assume that (a. b) = 1 and n = ak = b. By Corollary 1.40,
b | ak implies b | k. Thus, k = bk

and so n = ak = abk

.
1.61 Prove, for any (possibly negative) integers a and b, that (b. a) = (ba. a).
Solution. If c is a common divisor of a and b, then c | a and c | b; hence,
c | ba, and c is a common divisor of a and ba. This does not yet show
that the two gcds are equal. However, if c

is a common divisor of a and

b a, then c

| b, for b = a +(b a). Hence, c

is a common divisor of a
and b. It now follows that the two gcds are equal.
1.62 If a > 0, prove that a(b. c) = (ab. ac). [One must assume that a > 0 lest
a(b. c) be negative.]
Solution. Let (b. c) = d. Clearly, ad is a common divisor of ab and ac.
Now let k be a common divisor of ab and ac. It sufces to prove that
k | ad. If sb +t c = d, then asb +at c = ad, and k | ad because it divides
each of the summands on the left. As we mentioned in the exercise, the
only reason to assume that a > 0 is to guarantee that a(b. c) be positive.
1.63 Prove that the following pseudocode implements the Euclidean algorithm.
Input: a, b
Output: d
d := b; s := a
WHILE s > 0 DO
rem := remainder after dividing d by s
d := s
s := rem
END WHILE
Solution. The idea is to show that a proof of the Euclidean algorithm can
be constructed by following the steps of the algorithm.
1.64 If F
n
denotes the nth term of the Fibonacci sequence 0. 1. 1. 2. 3. 5. 8. . . .,
prove, for all n 1, that F
n+1
and F
n
are relatively prime.
Solution. The hint refers to the fact, which is the key step in antanairesis,
that (a. b) = (a b. b) whenever a > b. The proof is by induction on
n 1. The base step n = 1 is true, for (F
2
. F
1
) = (1. 1) = 1. For the
26
inductive step, use antanairesis and the dening recurrence,
(F
n+2
. F
n+1
) = (F
n+1
F
n
. F
n+1
)
= (F
n
. F
n+1
) = 1.
Here is a proof that is a variation of the same idea. Let n 1 be the
smallest integer for which F
n+1
and F
n
have gcd d > 1. We note that
n > 1 because (F
2
. F
1
) = (1. 1) = 1, and so n 1 1. But if d is a
common divisor of F
n+1
and F
n
, then d divides F
n1
= F
n+1
F
n
, so
that (F
n
. F
n1
) = 1. This contradicts n being the smallest index for which
(F
n+1
. F
n
) = 1.
1.65 (i) Show that if d is the greatest common divisor of a
1
. a
2
. . . . . a
n
,
then d =

t
i
a
i
, where t
i
is in Z for all i with 1 i n.
Solution. The set I of all linear combinations

t
i
a
i
of a
1
, a
2
. . . . ,
a
n
, where t
i
is in Z for 1 i n, satises the conditions of Corol-
lary 1.37. If d is the smallest positive element in I , then the proof
of Theorem 1.35 can be modied to show that d is the gcd.
(ii) Prove that if c is a common divisor of a
1
. a
2
. . . . . a
n
, then c | d.
Solution. The proof of Corollary 1.40 generalizes easily.
1.66 (i) Show that (a. b. c), the gcd of a, b, c, is equal to (a. (b. c)).
Solution. It sufces to prove that any common divisor of a. b. c
is a common divisor of a and (b. c), and conversely. But each of
these statements is easy to prove.
(ii) Compute (120, 168, 328).
Solution.
(120. 168. 328) = (120. (328. 168)) = (120. 8) = 8
1.67 (i) Consider a complex number z = q +i p, where q > p are positive
integers. Prove that
(q
2
p
2
. 2qp. q
2
+ p
2
)
is a Pythagorean triple by showing that |z
2
| = |z|
2
.
Solution. If z = q + i p, then |z
2
| = |z|
2
, by part (i). Now
z
2
= (q
2
p
2
) +i 2qp, so that |z
2
| = (q
2
p
2
)
2
+ (2qp)
2
. On
the other hand, |z|
2
= (q
2
+p
2
)
2
. Thus, if we dene a = q
2
p
2
,
b = 2qp, and c = q
2
+ p
2
, then a
2
+ b
2
= c
2
and (a. b. c) is a
Pythagorean triple.
(ii) Show that the Pythagorean triple (9. 12. 15) (which is not primi-
tive) is not of the type given in part (i).
27
Solution. Suppose there are q and p for (9. 12. 15). Then 2qp =
12 and qp = 6. Since q > p are positive integers, the only
possibilities are q = 6 and p = 1 or q = 3 and p = 2. The
rst possibility gives the Pythagorean triple (12. 35. 37) while the
second gives the Pythagorean triple (5. 12. 13).
(iii) Using a calculator which can nd square roots but which can dis-
play only 8 digits, show that
(19597501. 28397460. 34503301)
is a Pythagorean triple by nding q and p.
Solution. If q and p exist, then we have
q
2
+ p
2
= 34503301
q
2
p
2
= 19597501.
Therefore, 2p
2
= 14905800 and p
2
= 7452900. Hence, p =
2730. Finally, 2qp = 28397460, and so q = 5201. Since we were
able to nd q and p, the original trio does form a Pythagorean
triple.
1.68 True or false with reasons.
(i) |2
19
3
12
| -
1
2
.
Solution. False.
(ii) If r = p
g
1
1
p
g
n
n
, where the p
i
are distinct primes and the g
i
are
integers, then r is an integer if and only if all the g
i
are nonnega-
tive.
Solution. True.
(iii) The least common multiple [2
3
3
2
5 7
2
. 3
3
5 13] = 2
3
3
5

5
2
7
2
13,45.
Solution. True.
(iv) If a and b are positive integers which are not relatively prime, then
there is a prime p with p | a and p | b.
Solution. True.
(v) If a and b are relatively prime, then (a
2
. b
2
) = 1.
Solution. True.
1.69 (i) Find gcd(210. 48) using factorizations into primes.
Solution. 210 = 2
1
3
1
5
1
7
1
and 48 = 2
4
3
1
5
0
7
0
, so that
(210. 48) = 2
1
3
1
5
0
7
0
= 6.
(ii) Find gcd(1234. 5678).
28
Solution. 1234 = 2 617 (the reader is expected to prove that 617
is prime, using

617 - 25) and 5678 = 2 17 167, so that the
gcd = 2.
1.70 (i) Prove that an integer m 2 is a perfect square if and only if each
of its prime factors occurs an even number of times.
Solution. If m = a
2
and a = p
e
1
1
p
e
n
n
, then m = p
2e
1
1
p
2e
n
n
.
Conversely, if m = p
2e
1
1
p
2e
n
n
, then m = a
2
, where a =
p
e
1
1
p
e
n
n
.
(ii) Prove that if m is a positive integer for which

m is rational, then
m is a perfect square. Conclude that if m is not a perfect square,
then

m is irrational.
Solution. Let m = p
e
1
1
p
e
n
n
. If m is not a perfect square, then
at least one of the e
i
is odd. If

m = a,b, then mb
2
= a
2
. The
exponent of p
i
on the left is odd while the exponent of p
i
on the
right is even, and this is a contradiction.
1.71 If a and b are positive integers with (a. b) = 1, and if ab is a square, prove
that both a and b are squares.
Solution. Since a and b are relatively prime, the sets of primes occurring in
the factorization of a and of b are disjoint. Hence, if ab is a square, then all
the exponents e
i
in ab = p
e
1
1
p
e
n
n
are even, and hence all the exponents
arising from the primes in a (or in b) are even as well. Therefore, both a
and b are perfect squares.
1.72 Let n = p
r
m, where p is a prime not dividing an integer m 1. Prove
that p
_
n
p
r
_
.
Solution. Write a =
_
n
p
r
_
. By Pascals formula:
a =
_
n
p
r
_
=
n!
( p
r
)!(n p
r
)!
.
Cancel the factor (n p
r
)! and cross-multiply, obtaining:
a( p
r
)! = n(n 1)(n 2) (n p
r
+1).
Thus, the factors on the right side, other than n = p
r
m, have the form
n i = p
r
m i , where 1 i p
r
1. Similarly, the factors in
( p
r
)!, other than p
r
itself, have the form p
r
i , for i in the same range:
1 i p
r
1.
If p
e
| p
r
m i , where e r and i 1, then p
r
m i = bp
e
; hence,
p
e
| i ; there is a factorization i = p
e
j . Therefore, p
r
i = p
e
( p
re
j ).
29
A similar argument shows that if p
e
| p
r
i for i 1, then p
e
| p
r
m i .
By the fundamental theorem of arithmetic, the total number of factors p
occurring on each side must be the same. Therefore, the total number of
ps dividing ap
r
must equal the total number of ps dividing p
r
m. Since
p m, the highest power of p dividing p
r
m is p
r
, and so the highest power
of p dividing ap
r
is p
r
; that is, p a =
_
p
r
m
p
r
_
=
_
n
p
r
_
, as desired.
1.73 (i) For all rationals a and b, prove that
ab
p
= a
p
b
p
and a +b
p
max{a
p
. b
p
}.
Solution. If a = p
e
p
e
1
1
p
e
n
n
and b = p
f
p
f
1
1
p
f
n
n
, then
ab = p
e+f
p
e
1
+f
1
1
p
e
n
+f
n
n
.
Hence
ab
p
= p
ef
= p
e
p
f
= a
p
b
p
.
Assume e f , so that f e and a
p
= max{a
p
. b
p
}.
a +b = p
e
p
e
1
1
p
e
n
n
+ p
f
p
f
1
1
p
f
n
n
= p
e
_
p
e
1
1
p
e
n
n
+ p
f e
p
f
1
1
p
f
n
n
_
.
If u = p
e
1
1
p
e
n
n
+p
f e
p
f
1
1
p
f
n
n
, then either u = 0 or u
p
=
p
0
= 1. In the rst case, a + b
p
= 0, and the result is true.
Otherwise,
a +b
p
= p
e
u
p
= a
p
u
p
a
p
= max{a
p
. b
p
}.
(ii) For all rationals a, b, prove
p
(a. b) 0 and
p
(a. b) = 0 if and
only if a = b.
Solution.
p
(a. b) 0 because cp 0 for all c. If a = b, then

p
(a. b) = a bp = 0p = 0; conversely, if
p
(a. b) = 0,
then a b = 0 because 0 is the only element c with c
p
= 0.
(iii) For all rationals a, b, prove that
p
(a. b) =
p
(b. a).
Solution.
p
(a. b) =
p
(b. a) because
cp = 1pc
p
= c
p
.
30
(iv) For all rationals a, b, c, prove
p
(a. b)
p
(a. c) +
p
(c. b).
Solution.
p
(a. b)
p
(a. c) +
p
(c. b) because

p
(a. b) = a b
p
= (a c) +(c b)
p
max{a c
p
. c b
p
}
a c
p
+c b
p
=
p
(a. c) +
p
(c. b).
(v) If a and b are integers and p
n
| (a b), then
p
(a. b) p
n
.
(Thus, a and b are close if a b is divisible by a large power
of p.)
Solution. If p
n
| a b, then a b = p
n
u, where u is an integer.
But u
p
1 for every integer u, so that

(
a. b) = a b
p
= p
n
u
p
= p
n

p
u
p
p
n
.
At this point, one could assign a project involving completions,
p-adic integers, and p-adic numbers.
1.74 Let a and b be in Z. Prove that if
p
(a. b) p
n
, then a and b have the
same rst n p-adic digits, d
0
. . . . . d
n1
.
Solution. This follows from the fact that p
n
| a if and only if the rst
p-adic digits d
0
. . . . . d
n1
are all 0.
1.75 Prove that an integer M 0 is the lcm of a
1
. a
2
. . . . . a
n
if and only if it
is a common multiple of a
1
. a
2
. . . . . a
n
which divides every other common
multiple.
Solution. Consider the set I of all the common multiples of a
1
, a
2
. . . ., a
n
.
It is easy to check that I satises the hypotheses of Corollary 1.37, so that
every number m in I is a multiple of d, where d is the smallest positive
element in I . Since each a
i
is in I , d is a common multiple; if m is any
common multiple, then m is in I and hence d | m.
Conversely, if d is a common multiple dividing every common multiple
m, then d |m|.
1.76 (i) Give another proof of Proposition 1.56, [a. b](a. b) = |ab|, with-
out using the Fundamental Theorem of Arithmetic.
Solution. The result is easy if either a = 0 or b = 0. Otherwise,
if (a. b) = d, then a = da

and b = db

, where (a

. b

) = 1. Now
ab,d is an integer, and it is a common multiple of a and b:
ab,d = a(b,d) = b(a,d).
Next, we show that if c is a common multiple, then ab,d divides
c. By hypothesis, c = am = b. Now c = am = da

m and
31
c = b = db

, so that a

m = b

. Thus, a

divides b

; as
(a

. b

) = 1, we have a

divides , by Corollary 1.40. Write =

a

k, and observe that

c = db

= db

k = (db

)(da

)k,d = [ab,d]k.
Therefore, ab,d = [a. b], and so [a. b](a. b) = ab.
(ii) Find [1371, 123].
Solution. [1371. 123] = 1371 123,(1371. 123). By the Eu-
clidean algorithm, (1371. 123) = 3, and so [1371. 123] = 56. 211.
1.77 True or false with reasons.
(i) If a and m are integers with m > 0, then a i mod m for some
integer i with 0 i m 1.
Solution. True.
(ii) If a. b and m are integers with m > 0, then a b mod m implies
(a +b)
m
a
m
+b
m
mod m.
Solution. False.
(iii) If a is an integer, then a
6
a mod 6.
Solution. False.
(iv) If a is an integer, then a
4
a mod 4.
Solution. False.
(v) 5263980007 is a perfect square.
Solution. False.
(vi) There is an integer n with n 1 mod 100 and n 4 mod 1000.
Solution. False.
(vii) There is an integer n with n 1 mod 100 and n 4 mod 1001.
Solution. True.
(viii) If p is a prime and m n mod p, then a
m
a
n
mod p for every
natural number a.
Solution. False.
1.78 Find all the integers x which are solutions to each of the following congru-
ences:
(i) 3x 2 mod 5.
Solution. x 4 mod 5.
(ii) 7x 4 mod 10.
Solution. x 12 mod 10.
32
(iii) 243x +17 101 mod 725.
Solution. The Euclidean algorithm gives 1 = 182 24361 725.
243x +17 101 mod 725 gives 243x 84 mod 725.
Hence x 182 84 = 15288 63 mod 725.
(iv) 4x +3 4 mod 5.
Solution. x 4 mod 5.
(v) 6x +3 4 mod 10.
Solution. 6x +3 4 mod 10 is the same problem as 6x 1 mod
10. There are no solutions. The candidates for x are all r with 0
r 9, and multiplying each of them by 6 never gives 1 mod 10.
(Of course, (6. 10) = 1.)
(vi) 6x +3 1 mod 10.
Solution. 6x +3 1 mod 10 is the same problem as 6x 8 mod
10. This congruence does have solutions. If 6x 8 = 10m, then
3x 4 = 5m, and so this gives a new congruence 3x 4 mod 5
or x 8 3 mod 5. Thus, x = . . . 2. 3. 8. 13. . . .; there are
two possible solutions mod10, namely, x 3 mod 10 and x
8 mod 10.
1.79 Let m be a positive integer, and let m

be an integer obtained from m by

rearranging its (decimal) digits (e.g., take m = 314159 and m

= 539114).
Prove that m m

is a multiple of 9.
Solution. By casting out 9s, a number is divisible by 9 if and only if the
sum of its (decimal) digits is divisible by 9. But m and m

have the same

set of digits, for one is just a permutation of the other, and so the sum of
their digits is the same. Hence, one is divisible by 9 if and only if the other
one is.
1.80 Prove that a positive integer n is divisible by 11 if and only if the alternating
sum of its digits is divisible by 11.
Solution. Since 10 1 mod 11,
a = d
k
10
k
+ +d
1
10 +d
0
d
k
(1)
k
+ d
1
+d
0
.
1.81 What is the remainder after dividing 10
100
by 7?
Solution. Use Corollary 1.67 after noting that 100 = 2 7
2
+2 (of course,
this says that 100 has 7-adic digits 202). Hence
10
100
3
100
3
4
= 81 4 mod 7.
33
1.82 (i) Prove that 10q +r is divisible by 7 if and only if q 2r is divisible
by 7.
Solution. If 10q + r 0 mod 7, then 15q + 5r 0 mod 7,
and so q 2r 0 mod 7. Conversely, if q 2r 0 mod 7,
then 3q 6r 0 mod 7, hence 3q + r 0 mod 7, and so
10q +r 0 mod 7.
(ii) Given an integer a with decimal digits d
k
d
k1
. . . d
0
, dene
a

= d
k
d
k1
d
1
2d
0
.
Show that a is divisible by 7 if and only if some one of a

, a

,
a

,. . . is divisible by 7.
Solution. If a = 10b + d
0
, then a

= b 2d
0
. By part (i),
a 0 mod 7 if and only if a

0 mod 7. Now repeat.

1.83 (i) Show that 1000 1 mod 7.
Solution. Dividing 1000 by 7 leaves remainder 6 1 mod 7.
(ii) Show that if a = r
0
+1000r
1
+1000
2
r
2
+ , then a is divisible
by 7 if and only if r
0
r
1
+r
2
is divisible by 7.
Solution. If a = r
0
+1000r
1
+1000
2
r
2
+ , then
a r
0
+(1)r
1
+(1)
2
r
2
+ = r
0
r
1
+r
2
mod 7.
Hence a is divisible by 7 if and only if r
0
r
1
+r
2
is divisible
by 7.
1.84 For a given positive integer m, nd all integers r with 0 - r - m such that
2r 0 mod m
Solution. The answer depends on the parity of m.
1.85 Prove that there are no integers x, y, and z such that x
2
+ y
2
+ z
2
= 999.
Solution. Now 999 7 mod 8. But no sum of three numbers, with repeti-
tions allowed, taken from {0. 1. 4}, adds up to 7. This is surely true if a 4
is not used, while if a 4 is used, then the largest sum one can get which is
under 8 is 6.
1.86 Prove that there is no perfect square a
2
whose last two digits are 35.
Solution. If a is a positive integer, then
a = d
0
+10d
1
+100d
2
+ +10
n
d
n
.
where 0 d
i
9 for all i . Therefore, a d
0
+ 10d
1
mod 100. In
particular, the last two digits of a are 35 if and only if a 35 mod 100.
Let b be a positive integer with b
2
35 mod 100. Now the last digit
of b must be 5 (otherwise the last digit of b
2
would not be 5), and so
34
b 5 mod 10. Hence, b = 5 + 10q. and so b
2
(5 + 10q)
2
mod 100.
But
(5 +10q)
2
= 25 +2 5 q +100q
2
25 mod 100.
and so the last two digits of b
2
are 25, not 35. Therefore, no such b exists.
(More is true. We have proved that if the last digit of a perfect square is
5, then its last two digits are 25.)
1.87 If x is an odd number not divisible by 3, prove that x
2
1 mod 24.
Solution. Here are two ways to proceed. The odd numbers - 24 not
divisible by 3 are 1, 5, 7, 11, 13, 17, 19, 23; square each mod 24.
Alternatively, Example 1.161 says that the squares mod 8 are 0, 1, and
4. Now x
2
1 is divisible by 24 if and only if it is divisible by 3 and by
8 (as 3 and 8 are relatively prime). If x is to be odd, then x 0 mod 3 or
x 2 mod 3; looking at x mod 8, the hypothesis eliminates those x with
x
2
0 mod 8 or x
2
4 mod 8.
1.88 Prove that if p is a prime and if a
2
1 mod p, then a 1 mod p.
Solution. By Euclids lemma, if p divides a
2
1 = (a +1)(a 1), then
p divides a +1 or p divides a 1; that is, a 1 mod p.
1.89 Consider the congruence ax b mod m when gcd(a. m) = d. Show that
ax b mod m has a solution if and only if d | b.
Solution. If x
0
is a solution of ax b mod m, then ax
0
b = my
for some integer y. Now a = da

and m = dm

, by hypothesis, and so
da

x
o
b = dm

y. It follows that d | b.
Conversely, suppose that b = db

. Then the congruence is

da

x db

mod dm

.
Note that (a

. m

) = (a,d. m,d) = 1, because d is the gcd (a. m). There-

fore, the congruence a

x b

mod m

has a solution, say, u, and hence du

is a solution of the original congruence.
1.90 Solve the congruence x
2
1 mod 21.
Solution. If x
2
1 mod 21, then 21 | (x
2
1); that is, 21 | (x +1)(x 1).
Hence, 3 | (x + 1)(x 1) and 7 | (x + 1)(x 1). By Euclids lemma,
x 1 mod 3 and x 1 mod 7. Thus, x is not a multiple of 3, while
the candidates from the other congruences are 1, 8, 15, and 6, 13, 20. Thus,
there are 4 solutions: [1]. [8]. [13], and [20].
1.91 Solve the simultaneous congruences:
(i) x 2 mod 5 and 3x 1 mod 8;
Solution. x 27 mod 40.
(ii) 3x 2 mod 5 and 2x 1 mod 3.
Solution. x 14 mod 15.
35
1.92 Find the smallest positive integer which leaves remainder 4, 3, 1 after
dividing by 5, 7, 9, respectively.
Solution. That the desired integer x satises three congruences:
x 4 mod 5; x 3 mod 7; x 1 mod 9.
By the Chinese remainder theorem, the rst two congruences give
x 24 mod 35.
Now use the Chinese remainder theorem for the system
x 24 mod 35
x 1 mod 9
(which is possible because (35. 9) = 1. We obtain x 199 mod 315.
Thus, 199 is the smallest such solution.
1.93 How many days are there between Akbal 13 and Muluc 8 in the Mayan
tzolkin calendar?
Solution. Akbal is month 3 and Muluc is month 9. If x is the intervening
number of days, then
x 13 8 mod 13
x 3 9 mod 20.
The Chinese remainder theorem gives x 174 mod 260 (and so the num-
ber of intervening days is either 174 or 86).
1.94 (i) Show that (a + b)
n
a
n
+ b
n
mod 2 for all a and b and for all
n 1.
Solution. If a is even, then a + b b mod 2 and (a + b)
n

b
n
a
n
+ b
n
mod 2 for all n; a similar argument holds if b is
even. If both a and b are odd, then a 1 b mod 2; hence,
a + b 1 + 1 0 mod 2 and (a + b)
n
0 mod 2, while
a
n
+b
n
1 +1 = 0 mod 2.
(ii) Show that (a +b)
2
a
2
+b
2
mod 3.
Solution. If a = 1 = b, then (a + b)
2
4 1 mod 3, while
a
2
+b
2
2 mod 3.
1.95 Solve the linear system
x 12 mod 25
x 2 mod 30.
Solution. Absent.
36
1.96 Let m. m

be positive integers, let d = (m. m

), and let b b

mod d.
Prove that any two solutions of the system
x b mod m
x b

mod m

are congruent mod , where = lcm{m. m

}.
Solution. Absent.
1.97 On a desert island, ve men and a monkey gather coconuts all day, then
sleep. The rst man awakens and decides to take his share. He divides
the coconuts into ve equal shares, with one coconut left over. He gives
the extra one to the monkey, hides his share, and goes to sleep. Later,
the second man awakens and takes his fth from the remaining pile; he too
nds one extra and gives it to the monkey. Each of the remaining three men
does likewise in turn. Find the minimum number of coconuts originally
present.
Solution. Here are the equations arising from the story. Let C be the
number of coconuts.
C = 5a +1 (1)
4a = 5b +1 (2)
4b = 5c +1 (3)
4c = 5d +1 (4)
4d = 5e +1 (5)
We work from the bottom up. Since 4d occurs in the last equation and 5d
occurs in equation (4) above it, rewrite the latter as
16c = 5 4d +4.
Hence,
16c = 5 (5e +1) +4 = 25e +9.
Now go up to the next equation (3), which we multiply by 16:
64b = 5(16c) +16 = 5(25e +9) +16 = 125e +61.
Go up again after multiplying by 64:
256a = 5(64b) +64 = 5(125e +61) +64 = 625e +369.
Finally, multiply equation (1) by 256 to get
256C = 5(256a) +256 = 5(625e +369) +256 = 3125e +2101.
37
Thus,
256C 2101 mod 3125.
Were the hint, Try 4 coconuts, not given, one would proceed to solve
this congruence, taking note of the fact that (256. 3125) = 1. But C = 4
is a solution of this congruence, and so C 4 mod 3125; that is, every
number of the form 3125k 4 is a solution. The minimum value for C is
thus 3121 coconuts.
1.98 A suspect said that he had spent the Easter holiday April 21, 1893, with
his ailing mother; Sherlock Holmes challenged his veracity at once. How
could the great detective have been so certain?
Solution. April 21, 1893, fell on Friday, and so this date could not have
been Easter Sunday.
1.99 How many times in 1900 did the rst day of a month fall on a Tuesday?
Solution. The year y = 1900 was not a leap year, and
g(y) [19,4] 38 34 1 mod 7.
We seek the number of solutions to 3 1 + j (m) + 1 mod 7; that is,
j (m) 1 mod 7. For the interval March through December, there is only
one such month, namely, August. But we still have to check January and
February, for they behave as if they occurred in 1899. Now g(1899)
0 mod 7, and the congruence is
3 1 + j (m) mod 7 or j (m) 2 mod 7.
As j (January) = 0 and j (February) = 3, neither of these months gives a
solution, and so only August 1, 1900, fell on a Tuesday.
1.100 On what day of the week did February 29, 1896 fall? Conclude from your
method of solution that no extra fuss is needed to nd leap days.
Solution. March 1, 1896, fell on a Sunday, and so February 29, 1896, fell
on a Saturday.
1.101 (i) Show that 1987 had three Friday 13s.
Solution. The following dates fell on Friday in 1986: February
13, March 13, and November 13.
(ii) Show, for any year y > 0, that g(y) g(y 1) = 1 or 2, where
g(y) = y + y,4 y,100 + y,400.
Solution. The following dates fell on Friday in 1986: February
13, March 13, and November 13.
(iii) Can there be a year with exactly one Friday 13?
Solution. Yes. For example, October 13, 1988, fell on a Friday,
but no other 13th of the month fell on Friday.
38
1.102 My Uncle Ben was born in Pogrebishte, a village near Kiev, and he claimed
that his birthday was February 29, 1900. I told him that this could not be,
for 1900 was not a leap year. Why was I wrong?
Solution. Even though 1900 was not a leap year in America, it was a leap
year in Russia, which did not adopt the Gregorian calendar until after the
Russian Revolution.
Exercises for Chapter 2
2.1 True or false with reasons.
(i) If S T and T X, then S X.
Solution. True.
(ii) Any two functions f : X Y and g: Y Z have a composite
f g: X Z.
Solution. False.
(iii) Any two functions f : X Y and g: Y Z have a composite
g f : X Z.
Solution. True.
(iv) For every set X, we have X = .
Solution. True.
(v) If f : X Y and j : im f Y is the inclusion, then there is a
surjection g: X im f with f = j g.
Solution. True.
(vi) If f : X Y is a function for which there is a function g: Y X
with f g = 1
Y
, then f is a bijection.
Solution. False.
(vii) The formula f (
a
b
) = (a + b)(a b) is a well-dened function
Q Z.
Solution. False.
(viii) If f : N N is given by f (n) = n +1 and g: N N is given
by g(n) = n
2
, then the composite g f is n n
2
(n +1).
Solution. False.
(ix) Complex conjugation z = a + i b z = a i b is a bijection
C C.
Solution. True.
39
2.2 If A and B are subsets of a set X, prove that A B = A B

, where
B

= X B is the complement of B.
Solution. This is one of the beginning set theory exercises that is so easy
it is difcult; the difculty is that the whole proof turns on the meaning of
the words and and not. For example, let us prove that AB A B

.
If x A B, then x A and x , B; hence, x A and x B

, and so
x A B

. The proof is completed by proving the reverse inclusion.

2.3 Let A and B be subsets of a set X. Prove the de Morgan laws
(A B)

= A

and (A B)

= A

.
where A

= X A denotes the complement of A.

Solution. Absent.
2.4 If A and B are subsets of a set X, dene their symmetric difference (see
Figure 2.5) by
A + B = (A B) (B A).
(i) Prove that A + B = (A B) (A B).
Solution. Absent.
(ii) Prove that A + A = .
Solution. Absent.
(iii) Prove that A + = A.
Solution. Absent.
(iv) Prove that A +(B +C) = (A + B) +C (see Figure 2.6).
Solution. Show that each of A + (B + C) and (A + B) + C is
described by Figure 2.6.
(v) Prove that A (B +C) = (A B) +(A C).
Solution. Absent.
A B
Figure 2.5 Symmetric Difference
A B
C
Figure 2.6 Associativity
40
2.5 Let A and B be sets, and let a A and b B. Dene their ordered pair
as follows:
(a. b) = {a. {a. b}}.
If a

A and b

B, prove that (a

. b

) = (a. b) if and only if a

= a and
b

= b.
Solution. The result is obviously true if a

= a and b

= b.
For the converse, assume that
{a{a. b}} = {a

{a

. b

}}
There are two cases:
a = a

and {a. b} = {a

. b

};
a = {a

. b

} and {a. b} = a

.
If a = a

, we have {a. b} = {a

. b

} = {a. b

}. Therefore,
{a. b} {a} = {a. b

} {a}.
If a = b, the left side is empty, hence the right side is also empty, and so
a = b

; therefore, b = b

. If a = b, the the left side is {b}, and so the right

side is nonempty and is equal to {b

}. Therefore, b = b

, as desired.
In the second case, a = {a

. b

} = {{a. b}b

}. Hence,
a {a. b}
and
{a. b} {{a. b}. b

} = a.
contradicting the axiom a x a being false. Therefore, this case cannot
occur.
2.6 Let L = {(x. x) : x R}; thus, L is the line in the plane which passes
through the origin and which makes an angle of 45

with the x-axis.

(i) If P = (a. b) is a point in the plane with a = b, prove that L is
the perpendicular bisector of the segment PP

having endpoints
P = (a. b) and P

= (b. a).
Solution. The slope of L is 1, and the slope of PP

is
(b a),(a b) = 1. Hence, the product of the slopes is 1,
and so L is perpendicular to the PP

. The midpoint of PP

is
M = (
1
2
(a +b).
1
2
(a +b)), which lies on L, and
| PM| =
_
[a
1
2
(a +b)]
2
+[b
1
2
(a +b)]
2
= |MP

|.
41
(ii) If f : R R is a bijection whose graph consists of certain points
(a. b) [of course, b = f (a)], prove that the graph of f
1
is
{(b. a) : (a. b) f }.
Solution. By denition, f
1
(b) = a if and only if b = f (a).
Hence, the graph of f
1
consists of all ordered pairs
(b. f
1
(b)) = (b. a) = ( f (a). a).
2.7 Let X and Y be sets, and let f : X Y be a function.
(i) If S is a subset of X, prove that the restriction f |S is equal to the
composite f i , where i : S X is the inclusion map.
Solution. Both f |S and f i have domain S and target Y. If s S,
then ( f i )(s) = f (s) = ( f |S)(s). Therefore, f |S = f i , by
Proposition 2.2.
(ii) If im f = A Y, prove that there exists a surjection f

: X A
with f = j f

, where j : A Y is the inclusion.
Solution. For each x X, dene f

(x) = f (x). Thus, f

differs
from f only in its target.
2.8 If f : X Y has an inverse g, show that g is a bijection.
Solution. We are told that f g = 1
Y
and g f = 1
X
. Therefore, g has
an inverse, namely, f , and so g is a bijection.
2.9 Show that if f : X Y is a bijection, then it has exactly one inverse.
Solution. Let g: Y X and h : Y X both be inverses of f . Then
h = h1
Y
= h( f g) = (h f )g = 1
X
g = g.
2.10 Show that f : R R, dened by f (x) = 3x + 5, is a bijection, and nd
its inverse.
Solution. The function g, dened by g(x) =
1
3
(x 5), is the inverse of f ,
and so f is a bijection. (Alternatively, one could prove that f is a bijection
by showing directly that it is injective and surjective.)
2.11 Determine whether f : QQ Q, given by
f (a,b. c,d) = (a +c),(b +d)
is a function.
Solution. f is not a function:
1
2
=
2
4
and
2
6
=
1
3
, but
f (
1
2
.
2
6
) =
3
8
=
3
7
= f (
2
4
.
1
3
).
42
2.12 Let X = {x
1
. . . . . x
m
} and Y = {y
1
. . . . . y
n
} be nite sets, where the x
i
are
distinct and the y
j
are distinct. Show that there is a bijection f : X Y if
and only if |X| = |Y|; that is, m = n.
Solution. The hint is essentially the solution. If f is a bijection, there
are m distinct elements f (x
1
). . . . , f (x
m
) in Y, and so m n; using the
bijection f
1
in place of f gives the reverse inequality n m.
2.13 (Pigeonhole Principle)
(i) If X and Y are nite sets with the same number of elements,
show that the following conditions are equivalent for a function
f : X Y:
(i) f is injective;
(ii) f is bijective;
(iii) f is surjective.
Solution. Assume that X and Y have n elements. If f is injec-
tive, then there is a bijection from X to im f Y. Exercise 2.12
gives | im f | = n. It follows that im f = Y, for there can be no
elements in Y outside of im f , lest Y have more than n elements.
Any bijection is surjective, and so it remains to show that if f is
surjective, then it is injective. If Y = {y
1
. . . . . y
n
}, then for each
i , there exists x
i
X with f (x
i
) = y
i
. Were f not injective, there
would be i and x X with x = x
i
and f (x) = f (x
i
). This gives
n +1 elements in X, a contradiction.
(ii) Suppose there are 11 pigeons, each sitting in some pigeonhole. If
there are only 10 pigeonholes, prove that there is a hole containing
more than one pigeon.
Solution. Suppose each hole has at most one pigeon in it. If P is
the set of pigeons and H is the set of holes, dene f : P H by
f : pigeon h, where h is the hole containing it. Since each hole
contains at most one pigeon, f ( p
1
) = f ( p
2
) implies p
1
= p
2
,
where p
1
. p
2
are pigeons. Thus, f is an injection. By part (1), f
is a bijection, giving the contradiction 11 = 10.
2.14 Let f : X Y and g: Y Z be functions.
(i) If both f and g are injective, prove that g f is injective.
Solution. If (g f )(x) = (g f )(x

), then g( f (x)) = g( f (x

)).
Since g is injective, f (x) = f (x

); since f is injective, x = x

.
Hence, g f is injective.
(ii) If both f and g are surjective, prove that g f is surjective.
43
Solution. Let z Z. Since g is surjective, there is y Y with
g(y) = z; since f is surjective, there is x X with f (x) = y. It
follows that (g f )(x) = g( f (x)) = g(y) = z, and so g f is
surjective.
(iii) If both f and g are bijective, prove that g f is bijective.
Solution. By the rst two parts, g f is both injective and surjec-
tive
(iv) If g f is a bijection, prove that f is an injection and g is a sur-
jection.
Solution. If h = (g f )
1
, then (hg) f = 1 and g( f h) = 1. By
Lemma 2.9, the rst equation gives f an injection while the sec-
ond equation gives g a surjection.
2.15 (i) If f : (,2. ,2) R is dened by a tan a, then f has an
inverse function g; indeed, g = arctan.
Solution. By calculus, arctan(tan a) = a and tan(arctan x) = x.
(ii) Show that each of arcsin x and arccos x is an inverse function (of
sin x and cos x, respectively) as dened in this section.
Solution. Each of the other inverse trig functions satises equa-
tions analogous to sin(arcsin x) = x and arcsin(sin x) = x, which
shows that they are inverse functions as dened in this section.
2.16 (i) Let f : X Y be a function, and let {S
i
: i I } be a family of
subsets of X. Prove that
f
_
_
i I
S
i
_
=
_
i I
f (S
i
).
Solution. Absent.
(ii) If S
1
and S
2
are subsets of a set X, and if f : X Y is a function,
prove that f (S
1
S
2
) f (S
1
) f (S
2
). Give an example in which
f (S
1
S
2
) = f (S
1
) f (S
2
).
Solution. Absent.
(iii) If S
1
and S
2
are subsets of a set X, and if f : X Y is an injec-
tion, prove that f (S
1
S
2
) = f (S
1
) f (S
2
).
Solution. Absent.
2.17 Let f : X Y be a function.
(i) If B
i
Y is a family of subsets of Y, prove that
f
1
_
_
i
B
i
_
=
_
i
f
1
(B
i
) and f
1
_
_
i
B
i
_
=
_
i
f
1
(B
i
).
Solution. Absent.
44
(ii) If B Y, prove that f
1
(B

) = f
1
(B)

, where B

denotes the
complement of B.
Solution. Absent.
2.18 Let f : X Y be a function. Dene a relation on X by x x

if f (x) =
f (x

). Prove that is an equivalence relation. If x X and f (x) = y, the

equivalence class [x] is denoted by f
1
(y); it is called the ber over y.
Solution. Absent.
2.19 Let X = {rock. paper. scissors}. Recall the game whose rules are: paper
dominates rock, rock dominates scissors, and scissors dominates paper.
Draw a subset of X X showing that domination is a relation on X.
Solution. Absent.
2.20 (i) Find the error in the following argument which claims to prove
that a symmetric and transitive relation R on a set X must be re-
exive; that is, R is an equivalence relation on X. If x X and
x R y, then symmetry gives y R x and transitivity gives x R x.
Solution. There may not exist y X with x y.
(ii) Give an example of a symmetric and transitive relation on the
closed unit interval X = [0. 1] which is not reexive.
Solution. Dene
R = {(x. x) : 0 x
1
2
}.
Now R is the identity on Y = [0.
1
2
], so that it is symmetric and transitive.
However, R does not contain the diagonal of the big square X X, and so
R is not a reexive relation on X. For example, 1 1.
2.21 True or false with reasons.
(i) The symmetric group on n letters is a set of n elements.
Solution. False.
(ii) If S
6
, then
n
= 1 for some n 1.
Solution. True.
(iii) If . S
n
, then is an abbreviation for .
Solution. True.
(iv) If . are cycles in S
n
, then = .
Solution. False.
(v) If . are r-cycles in S
n
, then is an r-cycle.
Solution. False.
(vi) If S
n
is an r-cycle, then
1
is an r-cycle for every S
n
.
Solution. True.
45
(vii) Every transposition is an even permutation.
Solution. False.
(viii) If a permutation is a product of 3 transpositions, then it cannot
be a product of 4 transpositions.
Solution. True.
(ix) If a permutation is a product of 3 transpositions, then it cannot
be a product of 5 transpositions.
Solution. False.
(x) If
1
=
1
, then = .
Solution. False.
2.22 Find sgn() and
1
, where
=
_
1 2 3 4 5 6 7 8 9
9 8 7 6 5 4 3 2 1
_
.
Solution. In cycle notation, = (19)(28)(37)(46). Thus, is even, being
the product of four transpositions. Moreover, being a product of disjoint
transpositions, =
1
.
2.23 If S
n
xes some j , where 1 j n (that is, ( j ) = j ), dene

S
X
, where X = {1. . . . .

j . . . . . n}, by

(i ) = (i ) for all i = j .
Prove that
sgn(

) = sgn().
Solution. One of the cycles in the complete factorization of is the 1-
cycle ( j ). Hence, if there are t cycles in the complete factorization of ,
then there are t 1 cycles in the complete factorization of

. Therefore,
sgn(

) = (1)
[n1][t 1]
= (1)
nt
= sgn().
2.24 (i) If 1 - r n, prove that there are
1
r
[n(n 1) (n r +1)]
r-cycles in S
n
.
Solution. In the notation (i
1
i
2
. . . i
r
), there are n choices for i
1
,
n 1 choices for i
2
, . . ., n (r 1) = n r + 1 choices for i
r
.
We conclude that there are n(n 1) (n r +1) such notations.
However, r such notations describe the same cycle:
(i
1
i
2
. . . i
r
) = (i
2
i
3
. . . i
1
) = = (i
r
i
1
. . . i
r1
).
Therefore, there are
1
r
[n(n 1) (n r +1)] r-cycles in S
n
.
46
(ii) If kr n, where 1 - r n, prove that the number of permuta-
tions S
n
, where is a product of k disjoint r-cycles, is
1
k!
1
r
k
[n(n 1) (n kr +1).]
Solution. Absent.
2.25 (i) If is an r-cycle, show that
r
= (1).
Solution. If = (i
0
. . . i
r1
), then the proof of Lemma 2.25(ii)
shows that
k
(i
0
) = i
k
, where the subscript is read mod r. Hence,

r
(i
0
) = i
0
. But the same is true if we choose the notation for
having any of the other i
j
as the rst entry.
(ii) If is an r-cycle, show that r is the least positive integer k such
that
k
= (1).
Solution. Use Proposition 2.24. If k - r, then
k
(i
0
) = i
k
= i
0
,
so that
k
= 1.
2.26 Show that an r-cycle is an even permutation if and only if r is odd.
Solution. In the proof of Proposition 2.35, we showed that any r-cycle
is a product of r 1 transpositions. The result now follows from Proposi-
tion 2.39, for sgn() = (1)
r1
= 1.
2.27 Given X = {1. 2. . . . . n}, let us call a permutation of X an adjacency if
it is a transposition of the form (i i + 1) for i - n. If i - j , prove that
(i j ) is a product of an odd number of adjacencies.
Solution. We prove the result by induction on j i 1. The base step
is clear, for then is already an adjacency, and so it is a product of 1
adjacency. For the inductive step, we have
= (i j ) = (i i +1)(i +1 j )(i i +1).
by Proposition 2.32, for j i 2 implies j = i + 1. By induction,
(i + 1 j ) is a product of an odd number of adjacencies, and so is also
such a product.
2.28 Dene f : {0. 1. 2. . . . . 10} {0. 1. 2. . . . . 10} by
f (n) = the remainder after dividing 4n
2
3n
7
by 11.
(i) Show that f is a permutation.
Solution. Here is the two-rowed notation for f :
_
0 1 2 3 4 5 6 7 8 9 10
0 1 6 9 5 3 10 2 8 4 7
_
.
It follows that f is a permutation. (The reader is expected to use
his knowledge of congruences to facilitate the calculations.)
47
(ii) Compute the parity of f .
Solution. f = (2 6 10 7)(3 9 4 5). Since 4-cycles are odd, f is
even.
(iii) Compute the inverse of f .
Solution. f
1
= (7 10 6 2)(5 4 9 3).
2.29 (i) Prove that is regular if and only if is a power of an n-cycle.
Solution. If = (a
1
a
2
a
k
)(b
1
b
2
b
k
) (c
1
c
2
c
k
) is a
product of disjoint k-cycles involving all the numbers between 1
and n, show that =
k
, where
= (a
1
b
1
z
1
a
2
b
2
z
2
. . . a
k
b
k
z
k
).
(ii) Prove that if is an r-cycle, then
k
is a product of (r. k) disjoint
cycles, each of length r,(r. k).
Solution. Absent.
(iii) If p is a prime, prove that every power of a p-cycle is either a
p-cycle or (1).
Solution. Absent.
(iv) How many regular permutations are there in S
5
? How many regu-
lar permutations are there in S
8
?
Solution. Absent.
2.30 (i) Prove that if and are (not necessarily disjoint) permutations
that commute, then ()
k
=
k

k
for all k 1.
Solution. We prove rst, by induction on k 1, that
k
=
k
.
The base step is true because and commute. For the inductive
step,

k+1
=
k

=
k
(inductive hypothesis)
=
k

=
k+1
.
We now prove the result by induction on k 1. The base step is
obviously true. For the inductive step,
()
k+1
= ()
k
=
k

k
(inductive hypothesis)
=
k

k
(proof above)
=
k+1

k+1
.
48
(ii) Give an example of two permutations and for which ()
2
=

2
.
Solution. There are many examples. One is = (1 2) and =
(1 3). Since both and are transpositions,
2
= (1) =
2
, and
so
2

2
= (1). On the other hand, = (1 3 2), and ()
2
=
(1 3 2)
2
= (1 2 3) = (1).
2.31 (i) Prove, for all i , that S
n
moves i if and only if
1
moves i .
Solution. Since is surjective, there is k with k = i . If k = i ,
then i = i and i = j = i , a contradiction; hence, k = i . But

1
i = k, and so
1
moves i . The converse follows by replacing
by
1
, for (
1
)
1
= .
(ii) Prove that if . S
n
are disjoint and if = (1), then = (1)
and = (1).
Solution. By (i), if and are disjoint, then
1
and are dis-
joint: if moves some i , then
1
must x i . But = (1)
implies
1
= , so that there can be no i moved by . There-
fore, = (1) = .
2.32 If n 2, prove that the number of even permutations in S
n
is
1
2
n!.
Solution. Let = (1 2), and dene f : A
n
O
n
, where A
n
is the set of
all even permutations in S
n
and O
n
is the set of all odd permutations, by
f : .
If is even, then is odd, so that the target of f is, indeed, O
n
. The
function f is a bijection, for its inverse is g: O
n
A
n
, which is given by
g: .
2.33 Give an example of , , S
5
, none of which is the identity (1), with
= and = , but with = .
Solution. Set = (1 2). = (3 4), and = (3 5). Then = ,
= , and = .
2.34 If n 3, show that if S
n
commutes with every S
n
, then = (1).
Solution. If = (1), then it moves some i ; say, i = j = i . There is
with j = j and i = k = i . Then i = j = j , while i = k = j
(for is injective, and so k = i implies k = i = j ).
2.35 Can the following 15-puzzle be won?
4 10 9 1
8 2 15 6
12 5 11 3
7 14 13 #
Solution. No, because the associated permutation is odd.
49
2.36 True or false with reasons.
(i) The function e: N N N, dened by e(m. n) = m
n
, is an
associative operation.
Solution. False.
(ii) Every group is abelian.
Solution. False.
(iii) The set of all positive real numbers is a group under multiplication.
Solution. True.
(iv) The set of all positive real numbers is a group under addition.
Solution. False.
(v) For all a. b G, where G is a group, aba
1
b
1
= 1.
Solution. False.
(vi) Every permutation of the vertices :
1
. :
2
. :
3
of an equilateral trian-
gle
3
is the restriction of a symmetry of
3
.
Solution. True.
(vii) Every permutation of the vertices :
1
. :
2
. :
3
. :
4
of a square
4
is
the restriction of a symmetry of
4
.
Solution. False.
(viii) If a. b G, where G is a group, then (ab)
n
= a
n
b
n
for all n N.
Solution. False.
(ix) Every innite group contains an element of innite order.
Solution. False.
(x) Complex conjugation permutes the roots of every polynomial hav-
ing real coefcients.
Solution. True.
2.37 If a
1
. a
2
. . . . a
n
are (not necessarily distinct) elements in a group G, prove
that
(a
1
a
2
a
n
)
1
= a
1
n
a
1
2
a
1
1
.
Solution. Absent.
2.38 (i) Compute the order, inverse, and parity of
= (1 2)(4 3)(1 3 5 4 2)(1 5)(1 3)(2 3).
Solution.
= (1 2)(4 3)(1 3 5 4 2)(1 5)(1 3)(2 3) = (1 5 4)(2 3).
50
Its order is 6, its inverse is (1 4 5)(2 3), and it is odd.
(ii) What are the respective orders of the permutations in Exercises 2.22
and 2.28?
Solution. 2 and 4.
2.39 (i) How many elements of order 2 are there in S
5
and in S
6
?
Solution. In S
5
, there are
1
2
(5 4) = 10 transpositions and
1
2
[
1
2
(5 4)
1
2
(3 2)] = 15
products of two disjoint transpositions (the extra factor
1
2
so that
(a b)(c d) = (c d)(a b) not be counted twice).
In S
6
, there are
1
2
(6 5) = 15 transpositions,
1
2
[
1
2
(6 5)
1
2
(4 3)] = 45
products of two disjoint transpositions, and
1
6
[
1
2
(6 5)
1
2
(4 3)
1
2
(2 1)] = 15
products of three disjoint transpositions.
(ii) How many elements of order 2 are there in S
n
?
Solution.
1
2
n(n 1) +
1
2!
[
1
2
n(n 1)
1
2
(n 2)(n 3)]
+
1
3!
[
1
2
n(n 1)
1
2
(n 2)(n 3)
1
2
(n 4)(n 5)] + .
2.40 Let y be a group element of order m; if m = dt for some d 1, prove that
y
t
has order d.
Solution. Let x = y
t
. Now x
d
= (y
t
)
d
= t
t d
= y
m
= 1, and so the order
k of x is a divisor of d, by Lemma 2.53. But if x has order 1, then x = 1
and so y
t
= 1. This contradicts m = pt being the smallest positive integer
with y
m
= 1. Therefore, x has order p.
2.41 Let G be a group and let a G have order dk, where d. k > 1. Prove that
if there is x G with x
d
= a, then the order of x is d
2
k. Conclude that the
order of x is larger than the order of a.
Solution. It is clear that x
d
2
k
= (x
d
)
dk
= a
dk
= 1; thus, the order of x is
a divisor of d
2
k. Now x
dk
= a
k
= 1 because k - dk. Therefore, x has
order d
2
k. because p being a prime implies that there is no divisor d of
d
2
k with dk - d - d
2
k.
51
2.42 Let G = GL(2. Q), and let A =
_
0 1
1 0
_
and B =
_
0 1
1 1
_
. Show that
A
4
= I = B
6
, but that (AB)
n
= I for all n > 0. Conclude that AB can
have innite order even though both factors A and B have nite order.
Solution.
AB =
_
1 1
0 1
_
and (AB)
n
=
_
1 n
0 1
_
.
2.43 (i) Prove, by induction on k 1, that
_
cos sin
sin cos
_
k
=
_
cos k sin k
sin k cos k
_
.
Solution. The proof is by induction on k. The base step is obvious.
For the inductive step, let
A =
_
cos sin
sin cos
_
.
Then A
k+1
= AA
k
, and matrix multiplication gives the desired
result if one uses the addition formulas for sine and cosine.
(ii) Find all the elements of nite order in SO(2. R), the special or-
thogonal group.
Solution. By part (i), a matrix
A =
_
cos sin
sin cos
_
.
has nite order if and only if cos k = 1 and sin k = 0; that is,
when k is an integral multiple of 2. Thus, A has nite order if
= 2,k for some nonzero integer k.
2.44 If G is a group in which x
2
= 1 for every x G, prove that G must be
abelian.
Solution. If x G, then x
2
= 1 implies x
1
= x. Let y G. Since
1 = (xy)
2
= xyxy, we have xy = y
1
x
1
= yx.
2.45 Let G be a nite group in which every element has a square root; that is,
for each x G, there exists y G with y
2
= x. Prove that every element
in G has a unique square root.
Solution. Consider the function f : G G dened by f (x) = x
2
.
2.46 If G is a group with an even number of elements, prove that the number of
elements in G of order 2 is odd. In particular, G must contain an element
of order 2.
Solution. G is the disjoint union of X and Y, where
X = {g G : g
1
= g} and Y = {g G : g
1
= g}.
52
Now |X| is even, for it is the disjoint union of two-point subsets consisting
of an element and its inverse. Since |G| is even, we have |Y| even. But
1 Y, and so there is at least one element g Y with g = 1 and g
2
= 1
(indeed, there are an odd number of such elements.)
2.47 What is the largest order of an element in S
n
, where n = 1. 2. . . . . 10?
Solution. Denote the largest order of an element in S
n
by j(n). There is
n 1 2 3 4 5 6 7 8 9 10
j(n) 1 2 3 4 6 6 12 15 20 24
no known formula for j(n), though its asymptotic behavior is known, by a
theorem of E. Landau.
2.48 The stochastic group Y(2. R) consists of all those matrices in GL(2. R)
whose column sums are 1; that is, Y(2. R) consists of all the nonsingular
matrices
_
a c
b d
_
with a +b = 1 = c +d.
Prove that the product of two stochastic matrices is again stochastic, and
that the inverse of a stochastic matrix is stochastic.
Solution. If A and B are stochastic matrices, then
AB =
_
a c
b d
_ _
n y
x z
_
=
_
an +cx ay +cz
bn +dx by +dz
_
.
Now the sum of the entries in the rst column of AB is
(an +cx) +(bn +dx) = (a +b)n +(c +d)x
= n + x = 1.
Similarly, the sum of the entries in the second column of AB is 1, and so
AB is stochastic.
2.49 Show that the symmetry group Y(C) of a circle C is innite.
Solution. Rotation by any angle lies in Y(C).
2.50 Prove that every element in a dihedral group D
2n
has a unique factorization
of the form a
i
b
j
, where 0 i - n and j = 0 or 1.
Solution. The cyclic subgroup a has order n, hence index 2, so that D
2n
is the disjoint union a ab.
2.51 Let e
1
= (1. 0) and e
2
= (0. 1), If is an isometry of the plane xing
O, let (e
1
) = (a. b), (e
2
) = (c. d), and let A =
_
a c
b d
_
. Prove that
det(A) = 1.
Solution. Absent.
2.52 True or false with reasons. Here, G is always a group.
(i) If H is a subgroup of K and K is a subgroup of G, then H is a
subgroup of G.
53
Solution. True.
(ii) G is a subgroup of itself.
Solution. True.
(iii) The empty set is a subgroup of G.
Solution. False.
(iv) If G is a nite group and m is a divisor of |G|, then G contains an
element of order m.
Solution. False.
(v) Every subgroup of S
n
has order dividing n!.
Solution. True.
(vi) If H is a subgroup of G, then the intersection of two (left) cosets
of H is a (left) coset of H.
Solution. False.
(vii) The intersection of two cyclic subgroups of G is a cyclic sub-
group.
Solution. True.
(viii) If X is a nite subset of G, then X is a nite subgroup.
Solution. False.
(ix) If X is an innite set, then
F = { S
X
: moves only nitely many elements of X}
is a subgroup of S
X
.
Solution. True.
(x) Every proper subgroup of S
3
is cyclic.
Solution. True.
(xi) Every proper subgroup of S
4
is cyclic.
Solution. False.
2.53 Let H be a subgroup of a nite group G, and let a
1
H. . . . . a
t
H be a list of
all the distinct cosets of H in G. Prove the following statements without
using the equivalence relation on G dened by a b if b
1
a H.
(i) Prove that each g G lies in the coset gH, and that gH = a
i
H
for some i . Conclude that G = a
1
H a
t
H.
Solution. Absent.
(ii) If a. b G and aH bH = , prove that aH = bH. Conclude
that if i = j , then a
i
H a
j
H = .
Solution. Absent.
54
2.54 (i) Dene the special linear group by
SL(2. R) = {A GL(2. R) : det(A) = 1}.
Prove that SL(2. R) is a subgroup of GL(2. R).
Solution. It sufces to show that if A. B SL(2. R), then so is
AB
1
; that is, if det(A) = 1 = det(B), then det(AB
1
) = 1.
Since det(UV) = det(U) det(V), it follows that
1 = det(E) = det(BB
1
) = det(B) det(B
1
).
Hence, det(AB
1
) = det(A) det(B
1
) = 1.
(ii) Prove that GL(2. Q) is a subgroup of GL(2. R).
Solution. Both the product of two matrices with rational entries
and the inverse of a matrix with rational entries have rational en-
tries.
2.55 Give an example of two subgroups H and K of a group G whose union
H K is not a subgroup of G.
Solution. If G = S
3
, H = (1 2), and K = (1 3), then H K is not a
subgroup of G, for (1 2)(1 3) = (1 3 2) , H K.
2.56 Let G be a nite group with subgroups H and K. If H K, prove that
[G : H] = [G : K][K : H].
Solution. If G is a nite group with subgroup H, then [G : H] = |G|,|H|.
Hence, if H K, then
[G : K][K : H] = (|G|,|K|) (|K|,|H|) = |G|,|H| = [G : H].
2.57 If H and K are subgroups of a group G and if |H| and |K| are relatively
prime, prove that H K = {1}.
Solution. By Lagranges theorem, |H K| is a divisor of |H| and a divisor
of |K|; that is, |H K| is a common divisor of |H| and |K|. But |H| and
|K| are relatively prime, so that |H K| = 1 and H K = {1}.
2.58 Prove that every innite group contains innitely many subgroups.
Solution. An innite group have only nitely many cyclic subgroups.
2.59 Let G be a group of order 4. Prove that either G is cyclic or x
2
= 1 for
every x G. Conclude, using Exercise 2.44, that G must be abelian.
Solution. If G has order 4, then the only possible orders of elements in G
are 1, 2, and 4. If there is an element of order 4, then G is cyclic with that
element as generator. Otherwise, every element has order 1 or 2, so that
x
2
= 1 for every x G.
55
2.60 (i) Prove that the stochastic group Y(2. R), the set of all nonsingular
2 2 matrices whose row sums are 1, is a subgroup of GL(2. R).
Solution. The proof is a straightforward calculation: note that the
formula for the inverse is needed. The students should be encour-
aged to try to show that Y(3. R) is a group.
(ii) Dene Y

(2. R) to be the set of all nonsingular doubly stochastic

matrices (all row sums are 1 and all column sums are 1). Prove
that Y

(2. R) is a subgroup of GL(2. R).

Solution. The doubly stochastic group Y

is a subgroup because
it is the intersection of the subgroups Y and Y

.
2.61 Let G be a nite group, and let S and T be (not necessarily distinct)
nonempty subsets. Prove that either G = ST or |G| |S| +|T|.
Solution. Absent.
2.62 (i) If {S
i
: i I } is a family of subgroups of a group G, prove that
an intersection of cosets
_
i I
x
i
S
i
is either empty or a coset of
_
i I
S
i
.
Solution. Absent.
(ii) (B. H. Neumann.) If a group G is the set-theoretic union of
nitely many cosets,
G = x
1
S
1
x
n
S
n
.
prove that at least one of the subgroups S
i
has nite index in G.
Solution. Use induction on the number of distinct subgroups S
i
.
2.63 (i) Show that a left coset of (1 2) in S
3
may not be equal to a right
coset of (1 2) in S
3
; that is, there is S
3
with (1 2) =
(1 2).
Solution. Absent.
(ii) Let G be a nite group and let H G be a subgroup. Prove that
the number of left cosets of H in G is equal to the number of right
cosets of H in G.
Solution. Consider aH Ha
1
.
2.64 True or false with reasons.
(i) If G and H are additive groups, then every homomorphism
f : G H satises f (x + y) = f (x) + f (y) for all x. y G.
Solution. True.
(ii) A function f : R R

is a homomorphism if and only if

f (x + y) = f (x) + f (y) for all x. y R.
Solution. False.
56
(iii) The inclusion Z R is a homomorphism of additive groups.
Solution. True.
(iv) The subgroup {0} of Z is isomorphic to the subgroup {(1)} of S
5
.
Solution. True.
(v) Any two nite groups of the same order are isomorphic.
Solution. False.
(vi) If p is a prime, any two groups of order p are isomorphic.
Solution. True.
(vii) The subgroup (1 2) is a normal subgroup of S
3
.
Solution. False.
(viii) The subgroup (1 2 3) is a normal subgroup of S
3
.
Solution. True.
(ix) If G is a group, then Z(G) = G if and only if G is abelian.
Solution. True.
(x) The 3-cycles (7 6 5) and (5 26 34) are conjugate in S
100
.
Solution. True.
2.65 If there is a bijection f : X Y (that is, if X and Y have the same number
of elements), prove that there is an isomorphism : S
X
S
Y
.
Solution. Dene : S
X
S
Y
by : f f
1
. It is routine to
check that is a homomorphism; it is an isomorphism, for its inverse is the
function f
1
f .
2.66 Let G be a group, let X be a set, and let : G X be a bijection. Prove
that there is an operation on X which makes X into a group such that
: G X is an isomorphism.
Solution. Absent.
2.67 (i) Prove that the composite of homomorphisms is itself a homomor-
phism.
Solution. If f : G H and g: H K are homomorphisms,
then
(g f )(ab) = g( f (ab)) = g( f af b)
= g( f a)g( f b) = (g f )(a)(g f )(b).
(ii) Prove that the inverse of an isomorphism is an isomorphism.
Solution. If f is an isomorphism, then f
1
(x) = a if and only
if x = f (a). Hence, if f
1
(y) = b, then f
1
(xy) = ab (since
f (ab) = f (a) f (b) = xy), and so f
1
(xy) = ab = f
1
(x) f
1
(y).
57
If f : G H and g: G K are isomorphisms, then g
1
: K
G is an isomorphism, by (ii), and so f g
1
: K H is an iso-
morphism.
(iii) Prove that isomorphism is an equivalence relation on any family
of groups.
Solution. Absent.
(iv) Prove that two groups that are isomorphic to a third group are
isomorphic to each other.
Solution. Let f : G L and g: H L be isomorphisms. Then
g
1
: L H is an isomorphism, by part (ii), and the composite
g
1
f : G H is an isomorphism, by part (i).
2.68 Prove that a group G is abelian if and only if the function f : G G,
given by f (a) = a
1
, is a homomorphism.
Solution. If a. b G, where G is abelian, then (ab)
1
= b
1
a
1
=
a
1
b
1
. Conversely, assume that (ab)
1
= a
1
b
1
for all a. b G. Then
aba
1
b
1
= ab(ab)
1
= 1.
so that ab = ba.
2.69 This exercise gives some invariants of a group G. Let f : G H be an
isomorphism.
(i) Prove that if a G has innite order, then so does f (a), and if a
has nite order n, then so does f (a). Conclude that if G has an
element of some order n and H does not, then G

= H.
Solution. If a G has nite order n, then a
n
= 1 and 1 =
f (a
n
) = f (a)
n
; moreover, if k - n, then a
k
= 1 implies f (a
k
) =
1, because f is injective. Hence, f (a) has order n.
If a G has innite order, then f (a) has innite order in H.
Otherwise, f (a) has nite order n for some n, and so the isomor-
phism f
1
: H G would send f (a) a which is not of order
n, contradicting the result just proven.
(ii) Prove that if G

= H, then, for every divisor k of |G|, both G and
H have the same number of elements of order k.
Solution. Assume there are t elements in G of order k, and s
elements of order k in H. By (ii), for each such x G, the element
f (x) H has order k, so that t s. The reverse inequality
follows by considering f
1
: H G.
2.70 (i) Show that every group G with |G| - 6 is abelian.
Solution. If |G| = 1, then G is abelian. If |G| = 2. 3, or 5, then
G has prime order, hence G is cyclic, hence abelian. If |G| = 4,
then Exercise 2.59 shows that G is abelian.
58
(ii) Find two nonisomorphic groups of order 6.
Solution. A cyclic group of order 6 is not congruent to S
3
, for the
former is abelian and S
3
is not.
2.71 Prove that a dihedral group of order 4 is isomorphic to V, the 4-group, and
a dihedral group of order 6 is isomorphic to S
3
.
Solution. Absent.
2.72 Prove that any two dihedral groups of order 2n are isomorphic.
Solution. Absent.
2.73 Dene a function f : S
n
GL(n. R) by f : P

, where P

is the
matrix obtained from the n n identity matrix I by permuting its columns
by (the matrix P

is called a permutation matrix). Prove that f is an

isomorphism from S
n
to a subgroup of GL(n. R).
Solution. Absent.
2.74 (i) Find a subgroup H S
4
with H

= V but with H = V.
Solution. An example is
H = {1. (1 2). (3 4). (1 2 3 4)}.
(ii) Prove that the subgroup H in part (i) is not a normal subgroup.
Solution. If H S
4
and H is an element of order 2, then H
must contain all the conjugates of ; that is, H must contain all the
elements in S
4
having the same cycle structure as . Now either
is a transposition or a product of two disjoint transpositions.
Now Example 2.29 shows that S
4
contains 6 transpositions, and
they cannot all t inside the group H, which has order 4. There
are 3 products of disjoint transpositions, but if all of them lie in
H, then together with (1), they constitute H, forcing H = V, a
contradiction. Therefore, H is not a normal subgroup of S
4
.
2.75 If G is a group and a. b G, prove that ab and ba have the same order.
Solution. ab and ba are conjugates [ba = b(ab)b
1
], and hence they have
the same order.
2.76 (i) If f : G H is a homomorphism and x G has order k, prove
that f (x) H has order m, where m | k.
Solution. Since x
n
= 1, we have f (x)
n
= 1, and so the result
follows from Lemma 2.53.
(ii) If f : G H is a homomorphism and if (|G|. |H|) = 1, prove
that f (x) = 1 for all x G.
Solution. If (|G|. |H|) = 1, then f (x) has order k, where k | |G|;
hence, (k. |H|) = 1. It follows that
f (x) = f (x)
sk+t |H|
= f (x)
sk
f (x)
t |H|
= 1.
59
2.77 (i) Prove that the special orthogonal group SO(2. R) is isomorphic to
the circle group S
1
.
Solution. Dene
: A =
_
cos sin
sin cos
_
(cos . sin ).
In Example 2.48(iii), the product of two special orthogonal matri-
ces is computed, and this shows that the function is a homomor-
phism. The inverse function sends (cos . sin ) to the matrix in
the denition of .
(ii) Prove that all the rotations of the plane about the origin form a
group under composition which is isomorphic to SO(2. R).
Solution. By Proposition 2.59, every isometry xing the origin
is a linear transformation. The usual isomorphism between linear
transformations and matrices (done in general in Chapter 4) exists
here. If c
1
= (1. 0). c
2
= (0. 1) is the standard basis of R
2
, then
corresponds to the 2 2 matrix whose rst column are the coor-
dinates of (c
1
) and whose second column are the coordinates of
(c
2
), as in Example 2.48(iii).
2.78 Let G be the additive group of all polynomials in x with coefcients in Z,
and let H be the multiplicative group of all positive rationals. Prove that
G

= H.
Solution. List the prime numbers p
0
= 2. p
1
= 3. p
2
= 5. . . ., and dene
(e
0
+e
1
x +e
2
x
2
+ +e
n
x
n
) = p
e
0
0
p
e
n
n
.
It is routine to check that is a homomorphism, and it is clearly surjective.
By Corollary 1.53, is injective.
(In more advanced courses, one proves that both groups are free abelian
groups of countably innite rank.)
2.79 Show that if H is a subgroup with bH = Hb = {hb : h H} for every
b G, then H must be a normal subgroup.
Solution. If h H and b G, then bh bH = Hb, so that bh = h

b for
some h H. Therefore, bhb
1
= h

H, and so H is a normal subgroup.

2.80 Prove that the intersection of any family of normal subgroups of a group G
is itself a normal subgroup of G.
Solution. Let {N
i
: i I } be a family of normal subgroups of a group G,
and let D =
_
i I
N
i
. Now D is a subgroup of G, by Exercise 2.55. We
must show that gdg
1
D whenever d D. But d got into D by being in
every N
i
. Since N
i
G, we have gdg
1
N
i
for every g G and i I .
Therefore, gdg
1
D, and so D G.
60
2.81 Dene W = (1 2)(3 4), the cyclic subgroup of S
4
generated by (1 2)(3 4).
Show that W is a normal subgroup of V, but that W is not a normal sub-
group of S
4
. Conclude that normality is not transitive: K H and H G
need not imply K G.
Solution. Since every subgroup of an abelian group is a normal subgroup,
W is a normal subgroup of V. However, W is not a normal subgroup of S
4
,
for conjugating (1 2)(3 4) by (1 3) gives
(1 3)(1 2)(3 4)(1 3) = (2 3)(1 4) , W.
2.82 Let G be a nite group written multiplicatively. Prove that if |G| is odd,
then every x G has a square root. Conclude, using Exercise 2.45, that
there exists exactly one g G with g
2
= x.
Solution. The function : G G, dened by (g) = g
2
, is a homo-
morphism because G is abelian. Now ker = {g G : g
2
= 1} = {1}
because G has odd order, hence has no elements of order 2. It follows that
is injective. By Exercise 2.13, the function is surjective; that is, for
each x G, there is g G with x = (g) = g
2
. We have shown that x
has a square root. This square root is unique, for if g
2
= x = h
2
for some
h G, then (g) = (h), and so g = h because is injective.
2.83 Give an example of a group G, a subgroup H G, and an element g G
with
[G : H] = 3 and g
3
, H.
Solution. Take G = S
3
, take H = (1 2), and take g = (1 3). We have
[S
3
: H] = 6,2 = 3, and g
3
= g , H.
2.84 Show that the center of GL(2. R) is the set of all scalar matrices
_
a 0
0 a
_
with
a = 0.
Solution. Suppose that
A =
_
a b
c d
_
is a nonsingular matrix with c = 0; we will show that A is not in the
center of GL(2. R); that is, there is some nonsingular matrix B that does
not commute with A. If
B =
_
n x
y z
_
.
then
AB =
_
a b
c d
_ _
n x
y z
_
=
_

cn +dy
_
.
while
BA =
_
n x
y z
_ _
a b
c d
_
=
_

ay +cz
_
.
61
If AB = BA, then cn + dy = ay + cz; that is, c(n z) = y(a d).
There are two cases to consider. If a = d, then dene
B =
_
0 1
1 1
_
.
The 2,1 entry of AB is d = a, while the 2,1 entry of BA is a+c. As c = 0,
we have AB = BA. If a = d, dene
B =
_
1 1
1 0
_
.
Now the 2,1 entry of AB is c +d, while the 2,1 entry of BA is c +a. Since
a = d, these entries are different, and so AB = BA in this case as well.
Therefore, A is not in the center of GL(2. R). A similar argument holds if
b = 0, (This result is generalized to n n matrices in Corollary 4.86. The
proof using linear transformations is much shorter and much simpler.)
2.85 Let = e
2i ,n
be a primitive nth root of unity, and dene A =
_
0
0
1
_
and B =
_
0 1
1 0
_
.
(i) Prove that A has order n and that B has order 2.
Solution. It is clear that A
n
= I = B
2
. If 1 k - n, then
A
k
= I , for the 1,1 entry of A
k
is
k
= 1.
(ii) Prove that BAB = A
1
.
Solution. One multiplies the matrices.
(iii) Prove that the matrices of the form A
i
and BA
i
, for 0 i - n,
form a multiplicative subgroup G GL(2. C).
Solution. Let X = {A
i
: 0 i - n} and Y = {BA
i
: 0 i -
N}. Matrix multiplication is associative, and it is easy to see that
the nite set G = X Y is closed under inverses: if M G,
then M
1
G. Thus, it sufces to prove that G is closed under
multiplication. There are four cases.
XX G.
A
i
A
j
= A
i +j
X G.
XY G.
A
i
BA
j
= BA
i
A
j
= BA
j i
X Y = G.
Y X G.
BA
i
A
j
= BA
i +j
Y G.
YY G. Using the XY case,
(BA
i
)(BA
j
) = B(A
i
BA
j
) = BBA
j i
= A
j i
X G.
62
It follows that |G| = 2n, for A has order n.
(iv) Prove that each matrix in G has a unique expression of the form
B
i
A
j
, where i = 0. 1 and 0 j - n. Conclude that |G| = 2n
and that G

= D
2n
.
Solution. Note rst that B , A, for every matrix in A is diag-
onal and B is not diagonal. Suppose that B
i
A
j
= B
k
A

, where
the exponents lie in the proper ranges. If i = j , then we may can-
cel to obtain A
j
= A

. Since A has order n, we have n | j .

But | j | - n, and so j = . If i = j , then b A, and we
have already observed that this is not so. It follows that G contains
exactly 2n elements.
2.86 Recall that the group of quaternions Q (dened in Example 2.98) consists
of the 8 matrices in GL(2. C),
Q = {I. A. A
2
. A
3
. B. BA. BA
2
. BA
3
}.
where A =
_
0 1
1 0
_
and B =
_
0 i
i 0
_
.
(i) Prove that Q is a nonabelian group with operation matrix multipli-
cation.
Solution. We merely organize the needed calculations. First,
show that Q is closed under multiplication in the same way as
D
2n
was shown to be closed in the previous exercise. Dene
X = {A
i
: 0 i - 4} and Y = {BA
i
: 0 i - 4}, and
show that XX Q, XY Q, etc. Second, the inverse of each
matrix M Q also lies in Q.
(ii) Prove that I is the only element in Q of order 2, and that all other
elements M = I satisfy M
2
= I .
Solution. Straightforward multiplication.
(iii) Show that Q has a unique subgroup of order 2, and it is the center
of Q.
Solution. A group of order 2 must be a cyclic group generated
by an element of order 2. It is shown, in part (i), that I is the
only element of order 2. It is clear that I Z(Q), for scalar
matrices commute with every matrix. On the other hand, for every
element M = I , there is N Q with MN = NM.
(iv) Prove that I is the center Z(Q).
Solution. For each M Q but not in I , there is a matrix
M

Q with MM

= M

M.
2.87 Prove that the quaternions Q and the dihedral group D
8
are nonisomorphic
groups of order 8.
63
Solution. We use Exercise 2.69: Q has exactly one element of order 2,
while D
8
has 5 elements of order 2.
2.88 If G is a nite group generated by two elements of order 2, prove that
G

= D
2n
for some n 2.
Solution. Absent.
2.89 (i) Prove that A
3
is the only subgroup of S
3
of order 3.
Solution. Absent.
(ii) Prove that A
4
is the only subgroup of S
4
of order 12. (In Exer-
cise 2.135, this will be generalized from S
4
and A
4
to S
n
and A
n
for all n 3.)
Solution. If H is a subgroup of order 12, then H is normal (it
has index 2), and so it contains all the conjugates of each of its
elements. It must contain 1. We count the number of conjugates of
the various types of permutations (each count uses Exercise 2.24:
(1 2) has 6 conjugates; (1 2 3) has 8 conjugates; (1 2 3 4) has 6
conjugates; (1 2)(3 4) has 3 conjugates. The only way to get 12
elements is 1 +3 +8; but this is A
4
.
2.90 (i) Let Abe the set of all 22 matrices of the form A =
_
a b
0 1
_
, where
a = 0. Prove that A is a subgroup of GL(2. R).
Solution. It is a routine calculation to show that if M. N A,
then MN A and M
1
A.
(ii) Prove that : Aff(1. R) A, dened by f A, is an isomor-
phism, where f (x) = ax +b [see Example 2.48(iv)].
Solution. It is clear that is a bijection, and it remains to prove
that is a homomorphism, hence an isomorphism. Let g
Aff(1. R) be given by g(x) = cx +d. Now
g( f (x)) = g(ax +b)
= c(ax +b) +d
= cax +(cb +d).
Therefore,
( f g) =
_
ca cb +d
0 1
_
=
_
c d
0 1
_ _
a b
0 1
_
= (g)( f ).
(iii) Prove that the stochastic group Y(2. R) [see Exercise 2.48] is
isomorphic to the afne group Aff(1. R) by showing that
: Y(2. R) A

= Aff(1. R), given by (M) = QMQ
1
, is
an isomorphism, where Q =
_
1 0
1 1
_
and Q
1
=
_
1 0
1 1
_
.
64
Solution. It is easy to see that is a homomorphism:
(MN) = QMNQ
1
= QMQ
1
QNQ
1
= (M)(N).
Next, if
M =
_
a b
c d
_
.
then a +c = 1 and b+d = 1. It follows that im A, for matrix
multiplication gives
_
1 0
1 1
_ _
a b
c d
_ _
1 0
1 1
_
=
_
a b b
0 1
_
(the bottom row of the last matrix is
(a +c) (b +d) b +d.
and this is 0 1 because M is stochastic.
Finally, we must show that is surjective. This is obvious from
the last calculation. If
A =
_
a b
0 1
_
A.
then A = (M), where
M =
_
a +b b
1 a b 1 b
_
.
2.91 Prove that the symmetry group Y(
n
), where
n
is a regular polygon with
n vertices, is isomorphic to a subgroup of S
n
.
Solution. Every isometry Y(
n
) permutes the n vertices X of
n
. The
function f : Y(
n
) S
X

= S
n
, dened by |X, is easily seen to be
an injective homomorphism.
2.92 An automorphism of a group G is an isomorphism G G.
(i) Prove that Aut(G), the set of all the automorphisms of a group G,
is a group under composition.
Solution. Absent.
(ii) Prove that : G Aut(G), dened by g
g
(conjugation
by g), is a homomorphism.
Solution. Absent.
65
(iii) Prove that ker = Z(G).
Solution. Absent.
(iv) Prove that im Aut(G).
Solution. Absent.
2.93 If G is a group, prove that Aut(G) = {1} if and only if |G| 2.
Solution. Instructors are cautioned: assume that G is nite. The exercise
is, in fact, true as stated, but it wants Zorns lemma at one stage (see the
following argument).
If there is a G with a , Z(G), then conjugation by a is a nontrivial
automorphism; therefore, G is abelian. The map x x is an automor-
phism of G; if it is trivial, then x = x for all x G. Thus, we may
assume that that G is an abelian group (which we now write additively)
with 2x = 0; that is, G is a vector space over F
2
(the eld with 2 ele-
ments). If G is nite-dimensional and dim(G) 2, then any nonsingular
matrix other than the identity corresponds to a nontrivial automorphism.
For example, if :
1
. :
2
. . . . . :
n
is a basis, then there is an automorphism
which interchanges :
1
and :
2
and which xes :
3
. . . . . :
n
. If G is innite,
this argument works as well, if one knows that G has a 2-dimensional di-
rect summand. Assuming Zorns lemma, this is true, but the poor reader is
not prepared for this.
2.94 If C is a nite cyclic group of order n, prove that | Aut(C)| = (n), where
(n) is the Euler -function.
Solution. Absent.
2.95 True or false with reasons.
(i) If [a] = [b] in I
m
, then a = b in Z.
Solution. False.
(ii) There is a homomorphism I
m
Z dened by [a] a.
Solution. False.
(iii) If a = b in Z, then [a] = [b] in I
m
.
Solution. True.
(iv) If G is a group and K G, then there is a homomorphism G
G,K having kernel K.
Solution. True.
(v) If G is a group and K G, then every homomorphism G G,K
has kernel K.
Solution. False.
(vi) Every quotient group of an abelian group is abelian.
Solution. True.
66
(vii) If G and H are abelian groups, then G H is an abelian group.
Solution. True.
(viii) If G and H are cyclic groups, then G H is a cyclic group.
Solution. False.
(ix) If every subgroup of a group G is a normal subgroup, then G is
abelian.
Solution. False.
(x) If G is a group, then {1} G and G,{1}

= G.
Solution. True.
2.96 Prove that U(I
9
)

= I
6
and U(I
15
)

= I
4
I
2
.
Solution. The 6 elements in U(I
9
) are 1, 2, 4, 5,7, 8, and they form a cyclic
group with 2 as a generator.
The 8 elements of U(I
15
) are 1, 2, 4, 7, 8, 11, 13, 14. Now 2 has order 4:
2 = 1. 2. 4. 8.
and 11 has order 2. The intersection of these two subgroups is {1} and
they generate all of U(I
1
5) (if S = 2. 11, then |S| 5; by Lagranges
theorem, |S| is a divisor of 8; hence, |S| = 8 and S = U(I
1
5).
2.97 (i) If H and K are groups, prove, without using the rst isomorphism
theorem, that H

= {(h. 1) : h H} and K

= {(1. k) : k K}
are normal subgroups of H K with H

= H

and K

= K

.
Solution. First we show that H

is a subgroup of H K. It
contains (1. 1), which is the identity of H K, it is closed under
multiplication [(h. 1)(h

. 1) = (hh

. 1) H

], and it contains the

inverse of any of its elements [(h. 1)
1
= (h
1
. 1)].
We now show that H

(H K). If (x. 1) H

(so that x H)
and (h. k) H K, then
(h. k)(x. 1)(h. k)
1
= (h. k)(x. 1)(h
1
. k
1
)
= (hxh
1
. 1) H

.
Finally, we show that H

= H

. Dene : H H

by (h) =
(h. 1). It is routine to check that is an isomorphism.
(ii) Prove that f : H (H K),K

, dened by f (h) = (h. 1)K

,
is an isomorphism without using the rst isomorphism theorem.
Solution. Dene g: H (H K),K

by
g(x) = (x. 1)K

.
67
where x H.
First, we show that g is a homomorphism. If x. y H, then
g(xy) = (xy. 1)K

. On the other hand, the denition of multipli-

cation in the quotient group (H K),K

gives
g(x)g(y) = (x. 1)K

(y. 1)K

= (xy. 1)K

= g(xy).
as desired.
Next, we show that g is an injection. Now if x H, we have
g(x) = (x. 1)K

= {(x. 1)(1. k) : k K} = {(x. k) : k K}.

It follows that if g(x) = g(y), then
{(x. k) : k K} = {(y. k) : k K}.
and this implies that x = y.
Finally, we showthat g is a surjection. If (x. k)K

(HK),K

,
then (x. k)K

= (x. 1)K

[recall Lemma 2.82(i): aH = bH if

and only if b
1
a H; here, (x. 1)
1
(x. k) = (1. k) K

].
Therefore, (x. k)K

= (x. 1)K

= g(x), and so g is surjective.

Therefore, g is an isomorphism.
(iii) Prove K

(H K) and (H K),K

= H using the rst
isomorphism theorem.
Solution. The function f : H K H, dened by f : (h. k)
h, is a homomorphism:
f ((h. k)(h

. k

)) = f ((hh

. kk

)) = hh

= f ((h. k)) f ((h

. k

)).
Now f is surjective, for if h H, then h = f ((h. 1)), and
ker f = {(h. k) H K : f ((h. k)) = 1}
= {(1. k) H K : k K}
= K

.
The rst isomorphism theorem now gives K

H K and (H
K),K

= H. (Of course, the function H K K, sending
(h. k) k, is a homomorphism with kernel H

, and this implies

that H

H K.
2.98 If G is a group and G,Z(G) is cyclic, where Z(G) denotes the center of
G, prove that G is abelian; that is, G = Z(G). Conclude that if G is not
abelian, then G,Z(G) is never cyclic.
68
Solution. Suppose, on the contrary, that G,Z(G) is cyclic, say, with gen-
erator aZ(G). Since G is not abelian, a , Z(G). On the other hand, we
now show that a commutes with every element in g G. By hypothesis,
there is an integer i and an element z Z(G) with g = a
i
z. Therefore,
ag = aa
i
z = a
i
az = a
i
za = ga.
and this contradicts a , Z(G).
2.99 Let G be a nite group, let p be a prime, and let H be a normal subgroup of
G. Prove that if both |H| and |G,H| are powers of p, then |G| is a power
of p.
Solution. If |H| = p
h
and |G,H| = p
m
, then |G| = |G,H||H| = p
h+m
.
2.100 Call a group G nitely generated if there is a nite subset X G with
G = X. Prove that every subgroup S of a nitely generated abelian
group G is itself nitely generated.
Solution. Use induction on n 1, where X = {a
1
. . . . . a
n
}. The inductive
step should consider the quotient group G,a
n+1
.
2.101 (i) Let : G H be a surjective homomorphism with ker = T.
Let H = X, and, for each x X, choose an element g
x
G
with (g
x
) = x. Prove that G is generated by T {g
x
: x X}.
Solution. Absent.
(ii) Let G be a group and let T G. If both T and G,T are nitely
generated, prove that G is nitely generated.
Solution. Absent.
2.102 Let A. B and C be groups, and let . and be homomorphisms with
= .
A

//

@
@
@
@
@
@
@
B

C
If is surjective, prove that ker = (ker ).
Solution. If b (ker ), then b = (a), where (a) = 1. Hence,
(b) = (a) = (a) = 1. Therefore, ker (ker ).
For the reverse inclusion, let b ker , so that (b) = 1. Since is
surjective, there is a A with b = (a). Now (a) = (a) = (b) =
1, so that a ker and b = (a) (ker(bet a).
2.103 Let A and B be groups, let A

A and B

B be normal subgroups, and

let : A B be a homomorphism with (A

) B

.
(i) Prove that there is a (well-dened) homomorphism

: A,A

B,B

given by

: aA

(a)B

.
69
Solution. Absent.
(ii) Prove that if is surjective, then

is surjective.
Solution. Absent.
(iii) Give an example in which is injective and

is not injective.
Solution. Absent.
2.104 (i) Prove that Q,Z(Q)

= V, where Q is the group of quaternions and
V is the four-group. Conclude that the quotient of a nonabelian
group by its center can be abelian.
Solution. In Exercise 2.86, we saw that Z(Q) = {E}, so that
Q,Z(Q) has order 4. It is also shown in that exercise that if M =
I , then M
2
= I . It follows that every nonidentity element
in Q,Z(Q) has order 2, and hence, Q,Z(Q)

= V (any bijection
: Q,Z(Q) V with (1) = 1 must be an isomorphism).
(ii) Prove that Q has no subgroup isomorphic to V. Conclude that the
quotient Q,Z(Q) is not isomorphic to a subgroup of Q.
Solution. Exercise 2.86 shows that Q has a unique element of
order 2, whereas V has 3 elements of order 2.
2.105 Let G be a nite group with K G. If (|K|. [G : K]) = 1, prove that K
is the unique subgroup of G having order |K|.
Solution. Assume that H G and |H| = |K|. Let : G G,K
be the natural map; that is, (a) = aK for all a G. If h H, then
(h) = hK G,K has order n, say; by Exercise 2.60, n | |H| = |K|.
But n | |G,K| = [G : K], by Lagranges theorem, and so n is a common
divisor of |K| and |G,K|. Since (|K|. [G : K]) = 1, however, it follows
that n = 1. Therefore, hK ker = K, and hence H K. As H and K
have the same order, H = K, as desired.
2.106 Let H and K be subgroups of a group G.
(i) Prove that HK is a subgroup of G if and only if HK = K H.
In particular, the condition holds if hk = kh for all h H and
k K.
Solution. If HK is a subgroup. then it is closed under multi-
plication. But, if h H and k K, then h. k HK and
so kh HK. Therefore, K H HK. For the reverse inclu-
sion, if hk HK, then h
1
. k
1
. (hk)
1
HK. But hk =
(h
1
k
1
)
1
= (k
1
)
1
(h
1
)
1
K H, and so HK K H.
Conversely, assume that HK = K H. We use Proposition 2.68 to
prove that HK is a subgroup: HK = , for it contains 1, and
so it sufces to show that if x. y HK, then xy
1
HK. Now
x = h

and y = hk, where h

. h H and k

. k K. Hence,
70
xy
1
= h

(hk)
1
= h

k
1
h
1
. But k

k
1
= k

K and
k

k
1
h
1
= k

h
1
= h
1
k
1
for h
1
H and k
1
K because
K H = HK. Therefore, xy
1
= (h

h
1
)k
1
HK, as desired.
(ii) If HK = K H and H K = {1}, prove that HK

= H K.
Solution. Absent.
2.107 Prove the converse of Lemma 2.112: if K is a subgroup of a group G, and
if every left coset aK is equal to a right coset Kb, then K G.
Solution. Absent.
2.108 Let G be a group and regard G G as the direct product of G with itself.
If the multiplication j: GG G is a group homomorphism, prove that
G must be abelian.
Solution. If a. b G, then (a. 1) and (1. b) commute in G G, for
(a. 1)(1. b) = (a. b) = (1. b)(a. 1).
Now j: (a. b) ab. Hence,
ab = j((a. b)
= j((b. 1))((1. a))
= j((b. 1))j((1. a))
= ba.
Therefore, G is abelian.
2.109 Generalize Theorem 2.128 as follows. Let G be a nite (additive) abelian
group of order mn, where (m. n) = 1. Dene
G
m
= {g G : order (g) | m} and G
n
= {h G : order (h) | n}.
(i) Prove that G
m
and G
n
are subgroups with G
m
G
n
= {0}.
Solution. If x G
m
G
n
, then the order of x is a common divisor
of m and n. As the gcd(m. n) = 1, the element x has order 1 and
so x = 1.
(ii) Prove that G = G
m
+ G
n
= {g +h : g G
m
and h G
n
}.
Solution. Since G is abelian, every subgroup is normal, and so
the second isomorphism theorem applies. As G
m
G
n
) = {0}, we
have G
m

= (G
m
+ G
n
),G
n
, so that |G
m
+ G
n
| = |G
M
||G
N
| =
mn = |G|. Therefore, G
m
+ G
n
= G.
(iii) Prove that G

= G
m
G
n
.
Solution. The result now follows from Proposition 2.127.
71
2.110 (i) Generalize Theorem 2.128 by proving that if the prime factoriza-
tion of an integer m is m = p
e
1
1
p
e
n
n
, then
I
m

= I
p
e
1
1
I
p
en
n
.
Solution. The proof is by induction on n 2, the base case being
Exercise 2.109. For the inductive step, dene m

= p
e
1
1
p
e
n1
n1
.
Since (m

. p
e
n
n
) = 1, the result now follows from Theorem 2.128.
(ii) Generalize Corollary 2.131 by proving that if the prime factoriza-
tion of an integer m is m = p
e
1
1
p
e
n
n
, then
U(I
m
)

= U(I
p
e
1
1
) U(I
p
en
n
).
Solution. As in part (i), this is a straightforward induction on
n 2.
2.111 (i) If p is a prime, prove that ( p
k
) = p
k
(1
1
p
).
Solution. Of the numbers between 1 and p
n
, every pth one is a
multiple of p. Thus, there are p
n1
multiples of p, and p
n
p
n1
of these numbers are prime to p. Hence, ( p
n
) = p
n
p
n1
=
p
n
(1 1,p).
(ii) If the distinct prime divisors of a positive integer h are
p
1
. p
2
. . . . . p
n
, prove that
(h) = h(1
1
p
1
)(1
1
p
2
) (1
1
p
n
).
Solution. Let h = p
e
1
1
p
e
t
t
be the prime factorization of h. By
Corollary 2.131,
(h) = ( p
e
1
1
) ( p
e
t
t
)
= p
e
1
1
(1 1,p
1
) p
e
t
t
(1 1,p
t
)
= h(1 1,p
1
)(1 1,p
2
) (1 1,p
t
).
2.112 Let p be an odd prime, and assume that a
i
i mod p for 1 i p 1.
Prove that there exist a pair of distinct integers i and j with
i a
i
j a
j
mod p.
Solution. Note rst that a
i
i mod p implies

a
i

i mod p
( p 1)! mod p
1 mod p.
72
the last congruence by Wilsons theorem. If the statement of the exercise
is false, then each k between 1 and p 1 is congruent mod p to some i a
i
,
and so

k = ( p 1)!

i
i a
i
mod p.
But

i a
i
=

a
i
[( p 1)!]
2
mod p.
It follows that

i a
i
(1)
2
= 1 mod p. Since p is odd, 1 1 mod p,
and this is a contradiction.
2.113 If G is a group and x. y G, dene their commutator to be xyx
1
y
1
, and
dene the commutator subgroup G

to be the subgroup generated by all the

commutators. (The product of two commutators need not be a commutator,
but the smallest instance of this occurs in a group of order 96.)
(i) Prove that G

G.
Solution. Absent.
(ii) Prove that G,G

is abelian.
Solution. Absent.
(iii) If : G A is a homomorphism, where A is an abelian group,
prove that G

ker . Conversely, if G

ker , prove that im

is abelian.
Solution. Absent.
(iv) If G

H G, prove that H G.
Solution. Absent.
2.114 True or false with reasons.
(i) Every group G is isomorphic to the symmetric group S
G
.
Solution. False.
(ii) Every group of order 4 is abelian.
Solution. True.
(iii) Every group of order 6 is abelian.
Solution. False.
(iv) If a group G acts on a set X, then X is a group.
Solution. False.
(v) If a group G acts on a set X, and if g. h G satisfy gx = hx for
some x X, then g = h.
Solution. False.
73
(vi) If a group G acts on a set X, and if x. y X, then there exists
g G with y = gx.
Solution. False.
(vii) If g G, where G is a nite group, then the number of conjugates
of g is a divisor of |G|.
Solution. True.
(viii) Every group of order 100 contains an element of order 5.
Solution. True.
(ix) Every group of order 100 contains an element of order 4.
Solution. False.
(x) Every group of order 5
8
contains a normal subgroup of order 5
6
.
Solution. True.
(xi) If G is a simple group of order p
n
, where p is a prime, then n = 1.
Solution. True.
(xii) The alternating group A
4
is simple.
Solution. False.
(xiii) The alternating group A
5
is simple.
Solution. True.
(xiv) The symmetric group S
6
is simple.
Solution. False.
2.115 Prove that every translation
a
S
G
, where
a
: g ag, is a regular
permutation.
Solution. Absent.
2.116 Prove that no pair of the following groups of order 8,
I
8
; I
4
I
2
; I
2
I
2
I
2
; D
8
; Q.
are isomorphic.
Solution. None of the rst three groups can be isomorphic to any of the
last two because the rst three are abelian while D
8
and Q are not abelian.
Now I
8

= I
4
I
2
and I
8

= I
2
I
2
I
2
because I
8
has an element of
order 8 and the other two groups do not. Similarly, I
4
I
2

= I
2
I
2
I
2
because I
4
I
2
has an element of order 4 and I
2
I
2
I
2
does not. Finally,
D
8

= Q because Q has only 1 element of order 2 while D

8
has more than
one element of order 2.
2.117 If p is a prime and G is a nite group in which every element has order a
power of p, prove that G is a p-group.
Solution. If q is a prime divisor of |G| with q = p, then Cauchys theorem
gives an element in G of order q, contrary to the hypothesis.
74
2.118 Prove that a nite p-group G is simple if and only if |G| = p.
Solution. Absent.
2.119 Show that S
4
has a subgroup isomorphic to D
8
.
Solution. In Example 2.139, we have seen that D
8
can be viewed as certain
permutations of the 4 vertices of a square.
2.120 Prove that S
4
,V

= S
3
.
Solution. S
4
,V is a group of order 24/4=6, hence Proposition 2.135 shows
that it is isomorphic to either S
3
or I
6
. But S
4
,Vis not abelian: for example,
(1 2)V(1 3)V = (1 3)V(1 2)V because
(1 2)(1 3)[(1 3)(1 2)]
1
= (1 2)(1 3)(1 2)(1 3) = (2 3) , V.
Therefore, S
4
,V

= S
3
.
2.121 (i) Prove that A
4

= D
12
.
Solution. A
4
has no element of order 6, while D
12
does have such
an element.
(ii) Prove that D
12

= S
3
I
2
.
Solution. We may suppose that D
12
= a. b, where a
6
= 1 = b
2
and bab = a
1
. We know that the subgroup a has order 6, hence
index 2, and so there are two cosets:
D
12
= a ba.
Thus, every element x D
12
has a unique factorization x = b
i
a
j
,
where i = 0. 1 and 0 j - 6. Dene H = a
2
. b; now H

= S
3
,
for it is a nonabelian group of order 6; note that HD
12
because it
has index 2. If we dene K = a
3
, then |K| = 2 and K D
12
: it
sufces to prove that aa
3
a
1
K (which is, of course, obvious)
and ba
3
b K; but ba
3
b = a
3
= a
3
K. It is plain that
H K = {1} and HK = D
12
, and so D
12

= H K

= S
3
I
2
,
by Proposition 2.127.
2.122 (i) If H is a subgroup of G and if x H, prove that
C
H
(x) = H C
G
(x).
Solution. C
G
(x) = {g G : gx = xg} while C
H
(x) = {g H :
gx = xg}. It follows easily that each of C
H
(x) and H C
G
(x)
contains the other.
(ii) If H is a subgroup of index 2 in a nite group G and if x H,
prove that either |x
H
| = |x
G
| or |x
H
| =
1
2
|x
G
|, where x
H
is the
conjugacy class of x in H.
75
Solution. Write C = C
G
(x). Since H has index 2, we have
H G, by Proposition 2.97, and so the second isomorphism the-
orem gives |CH||C H| = |C||H|; by Lagranges theorem,
[H : C H] = [CH : C].
The left side is |x
H
|, for C H = C
H
(x), by part (i). But 2 =
[G : H] = [G : CH][CH : H] implies [CH : H] = 1 or
[CH : H] = 2, and this gives the result.
2.123 Prove that the group UT(3. I
2
) in Example 2.150 is isomorphic to D
8
.
Solution. You may use the fact that the only nonabelian groups of order 8
are D
8
and Q.
2.124 (i) How many permutations in S
5
commute with (1 2)(3 4), and how
many even permutations in S
5
commute with (1 2)(3 4).
Solution. If commutes with = (1 2)(3 4), then (5) = 5.
Hence, the problem has been reduced to looking within S
4
. Since
we cannot use centralizers and conjugates, there is no choice but
to look at the list of 24 elements of S
4
, checking which commute
with (1 2)(3 4). The answer is:
(1). (1 2)(3 4). (1 3)(2 4). (1 4)(2 3).
(1 2). (3 4). (1 3 24). (1 4 2 3).
(Note that if = (1 3 2 4), then
2
= .) Only the rst four of
these are even.
(ii) How many permutations in S
7
commute with (1 2)(3 4 5)?
Solution. There are
1
2
7 6
1
3
5 4 3 = 230 conjugates of in
S
7
, because two permutations in a symmetric group are conjugate
if and only if they have the same cycle structure. Now S
7
acts on
itself by conjugation; the orbit of is its conjugacy class, and its
stabilizer is its centralizer C
S
7
(). By the orbit-stabilizer theorem,
[S
7
: C
S
7
()] = 230. Therefore, |C
S
7
()| = 7!,230 = 12.
(iii) Exhibit all the permutations in S
7
commuting with (1 2)(3 4 5).
76
Solution. The following permutations commute with (1 2)(3 4 5):
(1) (6 7)
(1 2)(3 4 5) (1 2)(3 4 5)(6 7)
(1 2) (1 2)(6 7)
(3 4 5) (1 2)(6 7)(3 4 5)(6 7)
(1 2)(3 5 4) (1 2)(3 5 4)(6 7)
(3 5 4) (3 5 4)(6 7)
Since there are only 12 permutations commuting with , this must
be all of them.
2.125 (i) Show that there are two conjugacy classes of 5-cycles in A
5
, each
of which has 12 elements.
Solution. The hint shows that |C
S
5
()| = 5. Since || = 5 and
C
S
5
(), we have = C
S
5
(). By (i),
C
A
5
() = A
5
C
S
5
() = A
5
= .
so that |C
A
5
()| = 5. Therefore, the number of conjugates of in
A
5
is 60,|C
A
5
()| = 60,5 = 12.
(ii) Prove that the conjugacy classes in A
5
have sizes 1, 12, 12, 15,
and 20.
Solution. There are exactly 4 cycle structures in A
5
: (1); (1 2 3);
(1 2 3 4 5); (1 2)(3 4). Using Example 2.30, these determine
conjugacy classes in S
5
of sizes 1, 20, 24, and 15, respectively. In
part (ii), we saw that the class of 5-cycles splits, in A
5
, into two
conjugacy classes of size 12. The centralizer C
S
5
(1 2 3) consists
of
(1). (1 2 3). (1 3 2). (4 5). (4 5)(1 2 3). (4 5)(1 3 2);
Only the rst 3 of these are even, and so |C
A
5
((1 2 3))| = 3. It
follows from Corollary 2.145 that the conjugacy class of (1 2 3)
in A
5
has the same size as in S
5
, namely, 20. Finally, (1 2)(3 4)
has 15 conjugates in S
5
. By part (ii), there must be 15 conjugates
in A
5
, for the other alternative,
15
2
, is obviously impossible.
(iii) Prove that every normal subgroup H of a group G is a union of
conjugacy classes of G, one of which is {1}.
Solution. It follows from Proposition 2.142 that a group G is a
disjoint union of its conjugacy classes. Since a normal subgroup
of G contains all the conjugates of its elements, it follows that H
is a union of conjugacy classes of G.
77
(iv) Use parts (ii) and (iii) to give a second proof of the simplicity
of A
5
.
Solution. Since H contains 1, the order of H is a sumof 1 together
with some of the numbers 12, 12, 15, and 20. The only such sum
that divides 60 is 60 itself.
2.126 If . S
5
, where is a 5-cycle and is a transposition, prove that
. = S
5
.
Solution. Absent.
2.127 (i) For all n 3, prove that every A
n
is a product of 3-cycles.
Solution. We show that (1 2 3) and (i j k) are conjugate in
A
n
(and thus that all 3-cycles are conjugate in A
n
). If these cy-
cles are not disjoint, then each xes all the symbols outside of
{1. 2. 3. i. j ]}, say, and the two 3-cycles lie in A

, the group of all

even permutations on these 5 symbols. Of course, A

= A
5
, and,
as in Lemma 2.155, (1 2 3) and (i j k) are conjugate in A

; a for-
tiori, they are conjugate in A
n
. If the cycles are disjoint, then we
have just seen that (1 2 3) is conjugate to (3 j k) and that (3 j k)
is conjugate to (i j k), so that (1 2 3) is conjugate to (i j k) in
this case as well.
(ii) Prove that if a normal subgroup H A
n
contains a 3-cycle, where
n 5, then H = A
n
.
Solution. A normal subgroup H containing a 3-cycle must con-
tain every conjugate of ; as all 3-cycles are conjugate, H con-
tains every 3- cycle. But A
n
is generated by the 3-cycles, and so
H = A
n
.
2.128 Prove that (A
10
)

= A
10
, where G

denotes the commutator subgroup of a

group G. (See Exercise 2.113.)
Solution. We know that A
10
is a simple group and that (A
10
)

A
10
. Now
(A
10
)

= {1} because A
10
is not abelian. Therefore, (A
10
)

= A
10
.
2.129 Prove that the only normal subgroups of S
4
are {(1)}, V, A
4
, and S
4
.
Solution. We use Proposition 2.32. Let H be a normal subgroup of S
4
and
let H with = (1). If is a transposition, then H contains every
transposition. As S
4
is generated by the transpositions, we have H =
S
4
. If is a 3-cycle, then H contains all 3-cycles, and so A
4
H, by
Exercise 2.127. If is a product of disjoint transpositions, then V H.
Finally, if is a 4-cycle, say, H contains all 4-cycles. But
(1 2 3 4)(1 2 4 3) = (2 3) H.
Therefore, every transposition is in H, and so H = S
4
. Thus, if H = {1},
then V H. If V - H, then H contains some that is not a product
78
of disjoint transpositions, and the analysis above shows that H = A
4
or
H = S
4
.
2.130 Let {i
1
. . . . . i
r
} {1. 2. . . . n}, and let
F = { A
n
: xes all i with i = i
1
. . . . . i
r
}.
Prove that F

= A
r
.
Solution. Absent.
2.131 Prove that A
5
is a group of order 60 having no subgroup of order 30.
Solution. If H A
5
has order 30, then it has index 2. By Proposi-
tion 2.97(ii), H is a normal subgroup of A
5
, and this contradicts the fact
that A
5
is a simple group.
2.132 Let X = {1. 2. 3. . . .} be the set of all positive integers, and let S
X
be the
symmetric group on X.
(i) Prove that F

= { S
X
: moves only nitely many n X}
is a subgroup of S
X
.
Solution. Absent.
(ii) Dene A

to be the subgroup of F

generated by the 3-cycles.

Prove that A

is an innite simple group.

Solution. Absent.
2.133 (i) Prove that if a simple group G has a subgroup of index n, then G
is isomorphic to a subgroup of S
n
.
Solution. If [G : H] = n, let X denote the family of all the cosets
of H in G. The representation of G on the cosets of H gives a
homomorphism : G S
X

= S
n
with ker H. Since ker
is a normal subgroup, and since the only normal subgroups of G
are {1} and G, it follows that ker = {1} and is an injection.
(ii) Prove that an innite simple group has no subgroups of nite index
n > 1.
Solution. If G is an innite simple group, and if G has a subgroup
H of nite index n > 1, then (i) shows that the innite group G is
isomorphic to a subgroup of the nite group S
n
, a contradiction.
2.134 Let G be a group with |G| = mp, where p is a prime and 1 - m - p.
Prove that G is not simple.
Solution. By Cauchys theorem, G has an element of order p, and the
(cyclic) subgroup H it generates has order p and index m. The represen-
tation of G on the cosets of H gives a homomorphism : G S
m
with
ker H. If G is simple, we must have ker = {1}, so that G is iso-
morphic to a subgroup of S
m
. By Lagranges theorem, |H| | |S
m
|; that is,
79
p | m!. By Euclids lemma, we must have p | k for some k m, and this
contradicts m - p. Therefore, G is not simple.
2.135 If n 3, prove that A
n
is the only subgroup of S
n
of order
1
2
n!.
Solution. The case n = 3 is very easy, and n = 4 has been done in
Exercise 2.89; therefore, we may assume that n 5. If H is a second such
subgroup, then H S
n
because it has index 2. By the second isomorphism
theorem, H A
n
A
n
. Since A
n
is simple, either H A
n
= A
n
or
H A
n
= {(1)}. In the rst case, A
n
H, so that A
n
= H because they
have the same order. In the second case, we contradict the product formula,
which now says that
1
4
(n!)
2
| A
n
||H| = | A
n
H| |S
n
| = n!.
Here is a second proof. If H S
n
has index 2, then Proposition 2.97(i)
says that
2
H for every S
n
. Since
2
is always an even permutation,
we have
2
H A
n
for all S
n
. However, if is a 3-cycle, then it
has order 3, and so =
4
= (
2
)
2
. Therefore, H contains every 3-cycle,
and so A
n
H, by Exercise 2.127(i). Since both of these subgroups have
the same order, they are equal.
2.136 Prove that A
6
has no subgroups of prime index.
Solution. If H A
6
has prime index, then [A
6
: H] = p, where p =
2. 3. 5. The representation
p
: A
6
S
p
must be an injection, because
A
6
is simple and ker
p
H is a normal subgroup of A
6
. This gives the
contradiction 360 = | A
6
| p!.
2.137 True or false with reasons.
(i) If a nite group G acts on a set X, then X must be nite.
Solution. Faslse.
(ii) If a group G acts on a nite set X, then G must be nite.
Solution. False.
(iii) If a group G acts on a set X, and if x. y X, then G
x

= G
y
.
Solution. False.
(iv) If a group G acts on a set X, and if x. y X lie in the same orbit,
then G
x

= G
y
.
Solution. True.
(v) If D
10
acts on a bracelet with 5 beads, then the cycle structure of
D
10
is (1), (1 2)(3 4), or (1 2 3 4 5).
Solution. True.
(vi) If D
10
acts on a bracelet with 5 beads, and if is the reection
about the axis passing through one bead and perpendicular to the
opposite side, then the cycle index of is x
1
x
2
2
.
80
Solution. True.
2.138 How many ags are there with n stripes each of which can be colored any
one of q given colors?
Solution. We use Proposition 2.164. Here, G = , where
= (1 n)(2 n 1) (k k +1)
if n = 2k, and
= (1 n)(2 n 1) (k 1 k +1)(k)
if n = 2k +1. Therefore,
P
G
(x
1
. . . . . x
n
) =
_
1
2
(x
n
1
+ x
k
2
) if n = 2k
1
2
(x
n
1
+ x
1
x
k
2
) if n = 2k +1.
Thus, P
G
(q. . . . . q) =
1
2
(q
n
+ q
k
) or
1
2
(q
n
+ q
k+1
), depending on the
parity of n. One can give a master formula:
P
G
(q. . . . . q) =
1
2
(q
n
+q
[n,2]
).
2.139 Let X be the squares in an n n grid, and let be a rotation by 90

. Dene
a chessboard to be a (q. G)-coloring, where the cyclic group G = of
order 4 is acting. Show that the number of chessboards is
1
4
_
q
n
2
+q
(n
2
+1),2
+2q
(n
2
+3),4
_
.
where x is the greatest integer in the number x.
Solution. The group G that acts here is a cyclic group of order 4, where
is (clockwise) rotation by 90

. As
n 0. 1. 2. or 3 mod 4.
we have n
2
0 or 1 mod 4. Now G is acting on the n
2
squares, and one
sees that is a product of n
2
,4 disjoint 4-cycles in the rst case, whereas it
is a similar product with a 1-cycle otherwise. Now
3
=
1
has the same
cycle structure as , while each 4-cycle splits into two disjoint 2-cycles in

2
. Thus, when n
2
0 mod 4,
P
G
(x
1
. . . . . x
n
) =
1
4
(x
n
1
n +2x
n
2
,4
4
+ x
n
2
,2
2
)
and
P
G
(q. . . . . q) =
1
4
(q
n
+2q
n
2
,4
+q
n
2
,2
).
There is a similar formula when n
2
1 mod 4, and both can be combined
into the formula using greatest integer notation.
81
2.140 Let X be a disk divided into n congruent circular sectors, and let be
a rotation by (360,n)

. Dene a roulette wheel to be a (q. G)-coloring,

where the cyclic group G = of order n is acting. Prove that if n = 6,
then there are
1
6
(2q +2q
2
+q
3
+q
6
) roulette wheels having 6 sectors.
Solution. The group here is G = of order 6 acting by rotations. Thus,
= (1 2 3 4 5 6).

2
= (1 3 5)(2 4 6).

3
= (1 4)(2 5)(3 6).

4
= (1 5 3)(2 4).

5
= (1 6 5 4 3 2).
Hence
P
G
(x
1
. . . . . x
6
) =
1
6
(x
6
1
+2x
6
+2x
2
3
+ x
3
2
)
and
P
G
(q. . . . . q) =
1
6
(q
6
+2q +2q
2
+q
3
).
2.141 Let X be the vertices of a regular n-gon, and let the dihedral group G =
D
2n
act (as the usual group of symmetries). Dene a bracelet to be a
(q. G)-coloring of a regular n-gon, and call each of its vertices a bead.
(i) How many bracelets are there having 5 beads, each of which can
be colored any one of q available colors?
Solution. Proceed for the pentagon as we did for the square in
Example 2.139. If {:
0
. :
1
. :
2
. :
3
. :
4
} are the vertices of a regular
pentagon
5
, then we may regard the symmetry group Y(
5
) as
a subgroup of S
5
, for each Y(
5
) is a permutation of the
vertices (see Figure 2.15). There are 10 elements: 5 rotations:
(1). (:
0
:
1
:
2
:
3
:
4
). (:
0
:
2
:
4
:
1
:
3
).
(:
0
:
3
:
1
:
4
:
2
). (:
0
:
4
:
3
:
2
:
1
);
5 reections:
(:
1
:
4
)(:
2
:
3
). (:
0
:
2
)(:
3
:
4
). (:
0
:
4
)(:
1
:
3
).
(:
0
:
1
)(:
2
:
4
). (:
0
:
3
)(:
1
:
2
).
(ii) How many bracelets are there having 6 beads, each of which can
be colored any one of q available colors?
Solution. If {:
0
. :
1
. :
2
. :
3
. :
4
. :
5
} are the verices of a regular
hexagon
6
, then we may regard the symmetry group Y(
6
) as
82
a subgroup of S
6
, for each Y(
6
) is a permutation of the
vertices (draw a hexagon similar to Figure 2.15). There are 12
elements: 6 rotations:
(1). (:
0
:
1
:
2
:
3
:
4
:
5
).
(:
0
:
2
:
4
)(:
1
:
3
:
5
). (:
0
:
3
)(:
1
:
4
)(:
2
:
5
).
(:
0
:
4
:
2
)(:
1
:
5
:
3
). (:
0
:
5
:
4
:
3
:
2
:
1
);
6 reections:
(:
1
:
5
)(:
2
:
4
). (:
0
:
2
)(:
3
:
5
).
(:
0
:
4
)(:
1
:
3
). (:
0
:
1
)(:
2
:
5
)(:
3
:
4
).
(:
0
:
3
)(:
1
:
2
)(:
4
:
5
). (:
0
:
5
)(:
1
:
4
)(:
2
:
3
).
(iii) How many bracelets are there with exactly 6 beads having 1 red
bead, 2 white beads, and 3 blue beads?
Solution. Absent.
Exercises for Chapter 3
3.1 True or false with reasons.
(i) The subset {r +s : r. s Q} is a subring of R.
Solution. False.
(ii) Every subring of a domain is a domain.
Solution. True.
(iii) The zero ring is a subring of Z.
Solution. False.
(iv) There are innitely many positive integers m for which I
m
is a
domain.
Solution. True.
(v) If S is a subring of a commutative ring R, then U(S) is a subgroup
of U(R).
Solution. True.
(vi) If S is a subring of a commutative ring R, then U(S) = U(R) S.
Solution. False.
(vii) If R is an innite commutative ring, then U(R) is innite.
Solution. False.
83
(viii) If X is an innite set, then the family of all nite subsets of X
forms a subring of the Boolean ring B(X).
Solution. True.
3.2 Prove that a commutative ring R has a unique one 1; that is, if e R
satises er = r for all r R, then e = 1.
Solution. Assume that er = r for all r R. In particular, e1 = 1. On the
other hand, the dening property of 1 gives e1 = e. Hence, 1 = e.
3.3 Let R be a commutative ring.
(i) Prove the additive cancellation law.
Solution. Absent.
(ii) Prove that every a R has a unique additive inverse: if a +b = 0
and a +c = 0, then b = c.
Solution. Absent.
(iii) If u R is a unit, prove that its inverse is unique: if ub = 1 and
uc = 1, then b = c.
Solution. Absent.
3.4 (i) Prove that subtraction in Z is not an associative operation.
Solution. In Z, (a b) c = a (b c) = a b +c as long as
c = 0.
(ii) Give an example of a commutative ring R in which subtraction is
associative.
Solution. If R = F
2
, subtraction is the same as addition, and so it
is associative.
3.5 Assume that S is a subset of a commutative ring R such that
(i) 1 S;
(ii) if a. b S, then a +b S;
(iii) if a. b S, then ab S.
(In contrast to the denition of subring, we are now assuming a + b S
instead of ab S.) Give an example of a commutative ring R containing
such a subset S which is not a subring of R.
Solution. Let R = Z. If S is the subset consisting of all the positive
integers, then S satises the new axioms, but it is not a subring because it
is not closed under subtraction: for example, S does not contain 1 2.
3.6 Find the multiplicative inverses of the nonzero elements in I
11
.
Solution. Absent.
84
3.7 (i) If X is a set, prove that the Boolean group B(X) with elements the
subsets of X and addition given by symmetric difference,
U + V = (U V) (V U), is a commutative ring if one
denes multiplication by UV = U V.
Solution. The subset X is one, for X U = U for every sub-
set U X; 1 = 0 because X is nonempty; it is easy to check
associativityU (V W) = (U V) Wand commutativity
U V = V U. The proof of distributivity is longer.
A(B +C) = A [(B C) (C B)]
= [A (B C)] [A (C B)]
= [A B C

] [A C B

].
On the other hand,
AB + AC = (A B) +(A C)
= [(A B) (A C)

] [(A C) (A B)

]
= [(A B) (A

)] [(A C) (A

)]
= [(A B A

) (A B C

)]
[(A C A

) (A C B

)]
= (A B C

) (A C B

).
because A A

= .
(ii) Prove that B(X) contains exactly one unit.
Solution. If A X is a unit in B(X), then there is B X with
AB = 1; that is, A B = X. But X = A B A implies that
A = X.
(iii) If Y X (that is, Y is a proper subset of X), show that the one in
B(Y) is distinct from the one in B(X). Conclude that B(Y) is not
a subring of B(X).
Solution. The unit in B(Y) is Y. which is distinct from the unit in
B(X), namely, X.
(iv) Prove that every element U B(X) satises U
2
= U.
Solution. U U = U.
3.8 (i) If R is a domain and a R satises a
2
= a, prove that either
a = 0 or a = 1.
Solution. If a
2
= a, then 0 = a
2
a = a(a 1). Since R is a
domain, either a = 0 or a 1 = 0.
(ii) Show that the commutative ring F(R) in Example 3.11(i) contains
elements f = 0. 1 with f
2
= f .
85
Solution. Choose f to be any nonconstant function all of whose
values are 0 or 1.
3.9 Find all the units in the commutative ring F(R) dened in Example 3.11(i).
Solution. We claim that f is a unit if and only if f (r) = 0 for every r R.
If f is a unit, there is g F(R) with f g = 1; that is, f (r)g(r) = 1 for all
r R, and so f (r) = 0 for all r R.
Conversely, if f (r) = 0 for all r R, dene g F(R) by g(r) =
1,f (r); then f g = 1 and f is a unit.
3.10 Generalize the construction of F(R) to a set X and an arbitrary commu-
tative ring R: let F(X. R) be the set of all functions from X to R, with
pointwise addition f +g: x f (x) +g(x) and pointwise multiplication
f g: x f (x)g(x) for x X.
(i) Show that F(X. R) is a commutative ring.
Solution. One proves that F(X. R) is a commutative ring just as
one proves that F(R) is a commutative ring, for the only aspect if
R used in the proof is that it is a commutative ring.
(ii) Show that if X has at least two elements, then F(X. R) is not a
domain.
Solution. Denote two elements in X by 0 and 1. Dene f : R R
by f (0) = 1, f (1) = 0, and f (r) = 0 otherwise; dene g: R
R by g(0) = 0, g(1) = 1, and g(r) = 0 otherwise. Then neither
f = 0 nor g = 0, but f g = 0.
(iii) If R is a commutative ring, denote F(R. R) by F(R):
F(R) = {all functions R R}.
Show that F(I
2
) has exactly four elements, and that f + f = 0
for every f F(I
2
).
Solution. There are exactly four functions from a 2-point set to
itself. If f F(I
2
), then
( f + f )(r) = f (r) + f (r) = 0
for all r I
2
.
3.11 (i) Prove that the commutative ring C is a domain.
Solution. If zn = 0 and z = a +i b = 0, then z z = a
2
+b
2
= 0,
and
_
z
zz
_
z = 1.
(ii) Prove that Z, Q, and R are domains.
Solution. Every subring of a domain is a domain.
86
(iii) Prove that the ring of Gaussian integers is a domain.
Solution. The set Gaussian integers form a subring of C, and
hence it is a domain.
3.12 Prove that the intersection of any family of subrings of a commutative ring
R is a subring of R.
Solution. Absent.
3.13 Prove that the only subring of Z is Z itself.
Solution. Every subring R of Z contains 1, hence 1 +1. 1 +1 +1, etc, so
that R contains all positive integers (one needs induction), and nally, R
contains the additive inverses of these, i.e., all negative integers, as well.
3.14 Let a and m be relatively prime integers. Prove that if sa + t m = 1 =
s

a +t

m, then s s

mod m.
Solution. The given equation gives the congruences as 1 mod m and
as

1 mod m. Nowa and m relatively prime, and Theorem1.69 says that

any two solutions to ax 1 mod m are congruent; that is, s s

mod m.
3.15 (i) Is R = {a +b

2 : a. b Z} a domain?
Solution. It sufces to show that R contains 1 and is closed under
addition and multiplication. Each of these is routine: for example,
(a +b

2)(c +d

2) = (ac +2bd) +(ad +bc)

2.
(ii) Is R = {
1
2
(a +b

2) : a. b Z} a domain?
Solution. R is not a subring of R, hence is not a domain, for
(
1
2
)
2
=
1
4
, R.
(iii) Using the fact that =
1
2
(1+

19) is a root of x
2
x +5, prove
that R = {a +b : a. b Z} is a domain.
Solution. It is clear that R contains 1, and one shows easily that
if a. a

A and b. b

B, then R contains a a

, b b

, a b,
and aa

. Write a = +

5, b =
1
2
( +

5, and b

=
1
2
( +

5. It is easy to see that ab A if and

have the
same parity, while ab B otherwise; in either case, ab R.
Finally, write
bb

=
_
1
2
( +

5
_ _
1
2
( +

5
_
=
1
4
_
( +5

) +

5(

)
_
.
Expand and substitute = 2p + 1,

= 2p

+ 1, = 2q + 1,
and

= 2q

+ 1 (for ,

, , and

are odd). After collecting

terms, one sees that both the constant term and the coefcient of
87

5 are even; moreover, the quotients obtained after dividing each

by 2 have the same parity. It follows that bb

R.
This example can be generalized by replacing 5 by any integer D
with D 1 mod 4; the ring R is a special case of the ring of
integers in a quadratic number eld of the form Q(

D).
3.16 Prove that the set of all C

-functions is a subring of F(R).

Solution. Clearly the constant function 1 is a C

-function; if f and g
are C

-functions, then f g is obviously a C

-function, while f g is a
C

-function, by Exercise 1.42. Therefore, the set of all C

-functions is a
subring of F(R).
3.17 True or false with reasons.
(i) Every eld is a domain.
Solution. True.
(ii) There is a nite eld with more than 10
100
elements.
Solution. True.
(iii) If R is a domain, then there is a unique eld containing R.
Solution. False.
(iv) Every commutative ring is a subring of some eld.
Solution. False.
(v) The subset R = Q[i ] = {a +bi : a. b Q} is a subeld of C.
Solution. True.
(vi) The prime eld of Q[i ] = {a +bi : a. b Q} is Q.
Solution. True.
(vii) If R = Q[

2], then Frac(R) = R.

Solution. False.
3.18 (i) If R is a commutative ring, dene the circle operation a b by
a b = a +b ab.
Prove that the circle operation is associative and that 0 a = a for
all a R.
Solution. It is easy to see that 0 a = a for all a R:
0 a = 0 +a 0 a = a.
Let us show that is associative.
(a b) c = (a +b ab) c
= a +b ab +c (a +b ab)c
= a +b +c ab ac bc +(ab)c;
88
On the other hand,
a (b c) = a (b +c bc)
= a +b +c bc a(b +c bc)
= a +b +c bc ab ac +a(bc).
(ii) Prove that a commutative ring R is a eld if and only if the set
R
#
= {r R : r = 1}
is an abelian group under the circle operation.
Solution. Assume that R is a eld. To show that R
#
is a group,
we must rst show that is an operation on R
#
; that is, if a = 1
and c = 1, then a c = 1. If, on the contrary,
1 = a c = a +c ac = a +c(1 a).
then 1 a = c(1 a). Since a = 1, we may cancel 1 a,
and this gives the contradiction c = 1. Part (i) shows that is
associative and that 0 acts as the identity. It remains to nd the
inverse of an element a = 1; that is, to nd b = 1 such that
0 = a b = a +b ab. Dene b = a(a 1)
1
.
Conversely, assume that R
#
is a group. It sufces to nd a (multi-
plicative) inverse for every a = 0. Now a = 0 implies a +1 = 1,
and so there exists b R
#
with (a +1) b = 0; that is,
0 = a +1 +b (a +1)b = a +1 +b ab b.
Hence, 1 = a ab = a(1 b). It follows that (1 b) = a
1
.
3.19 (R. A. Dean) Dene F
4
to be the set of all 2 2 matrices
F
4
=
__
a b
b a +b
_
: a. b F
2
_
.
(i) Prove that F
4
is a commutative ring whose operations are matrix
addition and matrix multiplication.
Solution. Clearly I F
4
, and it is easy to check that F
4
is closed
under matrix addition:
_
a b
b a +b
_
+
_
c d
d c +d
_
=
_
a +c b +d
b +d a +c +b +d
_
.
Let us see that F
4
is closed under matrix multiplication:
_
a b
b a +b
_ _
c d
d c +d
_
=
_
ac +bd ad +bc +bd
bc +ad +bd bd +(a +b)(c +d)
_
.
89
Now so that the product lies in F
4
. It also follows from this for-
mula that multiplication in F
4
is commutative. All the other ax-
ioms of a commutative ring hold in F
4
because they hold in the full
matrix ring(for the cogniscenti, every subring of a (not necessarily
commutative) ring is a ring).
(ii) Prove that F
4
is a eld having exactly 4 elements.
Solution. If a matrix A =
_
a b
b a+b
_
in I
4
is nonzero, then a = 0 or
b = 0. Thus,
det(A) =
_

_
a = 0 if b = 0;
b = 0 if a = 0;
1 if a = 1 = b.
Thus, if A = 0, then det(A) = 1, and so the matrix A
1
exists.
As usual, A
1
=
_
a+b b
b a
_
, and so it is only a question of whether
this matrix lies in I
4
; that is, is the 2. 2 entry the sum of the entries
in row 1? The answer is yes, for (a +b) +b = a.
(iii) Show that I
4
is not a eld.
Solution. The commutative ring I
4
is not even a domain, for [2] =
[0] and [2][2] = [4] = [0].
3.20 Prove that every domain R with a nite number of elements must be a eld.
Solution. Let R

denote the set of nonzero elements of R. The cancella-

tion law can be restated: for each r R

, the function j
r
: R

,
dened by j
r
: x r x, is an injection R

. Since R

is nite,
Exercise 2.13 shows that every j
r
must also be a surjection. Hence, there
is s R

with 1 = j
r
(s) = rs, and so r has an inverse.
3.21 Find all the units in the ring Z[i ] of Gaussian integers.
Solution. If z = a +i b has an inverse, then there is u Z[i ] with zu = 1.
By Corollary 1.23, 1 = |zu| = |z||u|. Here, both |z| and |u| are integers,
so that |z| = 1. But |z| = a
2
+ b
2
, where a. b Z. Therefore, either
a = 1 and b = 0, or a = 0 and b = 1. That is, there are only four
units: 1. 1. i , and i .
3.22 Show that F = {a +b

2 : a. b Q} is a eld.
Solution. It is straightforward to check that
F = {a +b

2 : a. b Q}
is a subring of R, and so F is a commutative ring. If a + b

2 = 0, then
its inverse is c
1
(a b

2) F, where c = a
2
2b
2
(note that c = 0, by
Proposition 1.43).
90
3.23 (i) Show that F = {a +bi : a. b Q} is a eld.
Solution. It is straightforward to check that F is a subring of C,
and so it is a commutative ring; it is a eld because the inverse of
a +bi is r
1
(a bi ) F, where r = a
2
+b
2
.
(ii) Show that every u F has a factorization u =
1
, where
. Z[i ]. (See Exercise 3.50.)
Solution. Write
a +bi = ( p,q) +(r,s)i = ( ps +qri )(qs)
1
.
where p. q. r. s Z.
3.24 If R is a commutative ring, dene a relation on R by a b if there is a
unit u R with b = ua.
(i) Prove that is an equivalence relation.
Solution. Absent.
(ii) If a b, prove that (a) = (b), where (a) = {ra : r R}.
Conversely, prove that if R is a domain, then (a) = (b) implies
a b.
Solution. Assume that a b, so there is u. : R with a = ub
and u: = 1. We claim that (a) = (b). If x (a), then x = ra =
rub (b), and so (a) (b). For the reverse inclusion, if y (b),
then y = sb = s:a (a). Hence (a) = (b).
Conversely, assume that (a) = (b). If a = 0, then b = 0 and
so a b. We now assume that a = 0. Since a (b), there is
r R with a = rb; since b (a), there is s R with b = sa.
Therefore, a = rb = rsa, and so (1 rs)a = 0. Since R is a
domain, 1 = rs and r is a unit. Therefore, a b.
3.25 If R is a domain, prove that there is no subeld K of Frac(R) such that
R K Frac(R).
Solution. Absent.
3.26 Let k be a eld, and let R be the subring
R =
_
n 1 : n Z
_
k.
(i) If F is a subeld of k, prove that R F.
Solution. Absent.
(ii) Prove that a subeld F of k is the prime eld of k if and only if it is
the smallest subeld of k containing R; that is, there is no subeld
F

with R F

F.
Solution. Absent.
91
(iii) If R is a subeld of k, prove that R is the prime eld of k.
Solution. Absent.
3.27 (i) Show that every subeld of C contains Q.
Solution. Every subeld of C contains 1, hence contains Z, and
hence contains Q, for it must contain the multiplicative inverse of
every nonzero integer.
(ii) Show that the prime eld of R is Q.
Solution. This follows from (i).
(iii) Show that the prime eld of C is Q.
Solution. This follows from (i).
3.28 (i) For any eld F, prove that Y(2. F)

= Aff(1. F), where Y(2. F)
denotes the stochastic group.
Solution. The only properties of R used in setting up the isomor-
phism in Example 2.48(iv) is that it is a eld.
(ii) If F is a nite eld with q elements, prove that
|Y(2. F)| = q(q 1).
Solution. Now Aff(1. F) consists of all functions f : x ax +b,
where a. b F and a = 0, there are q 1 choices for a and q
choices for b, and so | Aff(1. F)| = q(q 1). But isomorphic
groups have the same order, and so |Y(2. F)| = q(q 1).
(iii) Prove that Y(2. F
3
)

= S
3
.
Solution. By part (ii), |Y(2. F
3
)| = 6. By Proposition 2.135, it is
isomorphic to S
3
or I
6
; as it is nonabelian, Y(2. F
3
)

= S
3
.
3.29 True or false with reasons.
(i) The sequence notation for x
3
2x +5 is (5. 2. 0. 1. 0. ).
Solution. True.
(ii) If R is a domain, then R[x] is a domain.
Solution. True.
(iii) Q[x] is a eld.
Solution. False.
(iv) If k is a eld, then the prime eld of k[x] is k.
Solution. True.
(v) If R is a domain and f (x). g(x) R[x] are nonzero, then deg( f g) =
deg( f ) +deg(g).
Solution. True.
92
(vi) If R is a domain and f (x). g(x) R[x] are nonzero, then either
f (x) + g(x) = 0 or deg( f + g) max{deg( f ). deg(g)}.
Solution. True.
(vii) If k is a eld, then k[x] = k(x).
Solution. False.
3.30 Show that if R is a nonzero commutative ring, then R[x] is never a eld.
Solution. If f (x) = x
1
, then x f (x) = 1. But the degree of the left side
is at least 1, while the degree of the right side is 0.
3.31 Let k be a eld and let A be an n n matrix with entries in k (so that the
powers A
i
are dened). If f (x) = c
0
+c
1
x + +c
m
x
m
k[x], dene
f (A) = c
0
I +c
1
A + +c
m
A
m
.
(i) Prove that k[A], dened by k[A] = { f (A) : f (x) k[x]}, is a
commutative ring under matrix addition and matrix multiplication.
Solution. Absent.
(ii) If f (x) = p(x)q(x) k[x] and if A is an n n matrix over k,
prove that f (A) = p(A)q(A).
Solution. Absent.
(iii) Give examples of n n matrices A and B such that k[A] is a
domain and k[B] is not a domain.
Solution. Absent.
3.32 (i) Let R be a domain. Prove that a polynomial f (x) is a unit in R[x]
if and only if f (x) is a nonzero constant which is a unit in R.
Solution. If f (x) is a unit, then there is g(x) R[x] with
f (x)g(x) = 1. Since R is a domain, deg( f ) + deg(g) = 0.
But degrees are nonnegative, so that deg( f ) = 0 and f (x) is a
nonzero constant.
If R is a eld, then every nonzero element is a unit in R; that is,
there is : R with u: = 1. Since R R[x], we have : R[x],
and so u is a unit in R[x].
(ii) Showthat ([2]x+[1])
2
= [1] in I
4
[x]. Conclude that the statement
in part (i) may be false for commutative rings that are not domains.
Solution. ([2]x +[1])
2
= [4]x
2
+[4]x +[1] = [1] in I
4
[x].
3.33 Show that if f (x) = x
p
x F
p
[x], then its polynomial function
f
.
: F
p
F
p
is identically zero.
Solution. Let f (x) = x
p
x F
p
[x]. If a F
p
, Fermats theorem gives
a
p
= a, and so f (a) = a
p
a = 0.
93
3.34 (i) If p is a prime and m. n N, prove that
_
pm
pn
_

_
m
n
_
mod p.
Solution. Absent.
(ii) Prove that
_
p
r
m
p
r
n
_

_
m
n
_
mod p for all r 0.
Solution. Absent.
(iii) Give another proof of Exercise 1.72: if p is a prime not dividing
an integer m 1, then p
_
p
r
m
p
r
_
.
Solution. Absent.
3.35 Let C, and let Z[] be the smallest subring of C containing ; that
is, Z[] =
_
S, where S ranges over all those subrings of C containing .
Prove that
Z[] = { f () : f (x) Z[x]}.
Solution. Absent.
3.36 Prove that the usual rules of calculus hold for derivatives of polynomials in
R[x], where R is a commutative ring; that is,
( f + g)

= f

+ g

;
(r f )

= r f

if r R;
( f g)

= f g

+ f

g;
( f
n
)

= nf
n1
f

for all n 1.
Solution. Let f (x) =

a
n
x
n
and g(x) =

b
n
x
n
.
(i)
( f + g)

=
_

a
n
x
n
+

b
n
x
n
_

=
_

(a
n
+b
n
)x
n
_

n(a
n
+b
n
)x
n1
=

na
n
x
n1
+

nb
n
x
n1
= f

+ g

.
(ii)
(r f )

=
_
r

a
n
x
n
_

=
_

ra
n
x
n
_

rna
n
x
n1
= r

na
n
x
n1
= r f

.
94
(iii) We prove that ( f g)

= f g

+ f

g by induction on deg( f ) (of course,
the equation holds when f = 0). The base step deg( f ) = 0 has been
done in part (ii). For the inductive step, write f (x) = a
k
x
k
+ h(x),
where h(x) = 0 or deg(h) - k = deg( f ). If h(x) = 0, then
( f g)

=
_
a
k
x
k

b
n
x
n
_

=
_

a
k
b
n
x
n+k
_

(n +k)a
k
b
n
x
n+k1
=

ka
k
b
n
x
n+k1
+

na
k
b
n
x
n+k1
= ka
k
x
k1

b
n
x
n
+a
k
x
k

nb
n
x
n1
= (a
k
x
k
)

g(x) +a
k
x
k
g

.
If h(x) = 0, then
( f g)

=
_
(a
k
x
k
+h(x))g(x)
_

=
_
a
k
x
k
g(x) +h(x)g(x)
_

=
_
a
k
x
k
g(x)
_
+ D(hg)

=
_
a
k
x
k
_

g(x) +a
k
x
k
g

+h

g +hg

=
_
a
k
x
k
+h
_

g +(a
k
x
k
+h)g

= f

g + f g

.
(iv) We prove that ( f
n
)

= nf
n1
f

by induction on n 1. If we dene
f
0
to be the constant function 1, then the base step holds. For the
inductive step,
_
f
n+1
_

=
_
f
n
f
_

=
_
f
n
_

f + f
n
f

= (nf
n1
f

) f + f
n
f

= (n +1) f
n
f

.
3.37 Assume that (x a) | f (x) in R[x]. Prove that (x a)
2
| f (x) if and only
if (x a) | f

in R[x].
95
Solution. Assume that x a | f (x). If (x a)
2
| f (x), then f (x) =
(x a)
2
h(x), and so f

(x) = 2(x a)h(x) + (x a)
2
h

(x). Hence,
x a | f

. (The assumption (x a) | f (x) is not needed in this direction.)
Conversely, assume that x a | f (x) and x a | f

; hence f (x) =
(x a)g(x) and f

(x) = (x a)k(x). Now
(x a)k(x) = f

(x) = ((x a)g)

= g +(x a)g

.
and so x a | g; that is, g(x) = (x a)(x). Hence, f (x) = (x a)g(x) =
(x a)
2
(x).
3.38 (i) If f (x) = ax
2p
+bx
p
+c F
p
[x], prove that f

(x) = 0.
Solution. If f (x) = ax
2p
+bx
p
+c F
p
[x], then
f (x) = 2pax
2p1
+ pbx
p1
= 0.
(ii) State and prove a necessary and sufcient condition that a polyno-
mial f (x) F
p
[x] have f

(x) = 0.
Solution. The condition is that there should be a polynomial
g(x) =

b
n
x
n
with f (x) = g(x
p
); that is, f (x) =

a
n
x
np
and b
p
n
= a
n
for all n. As in part (i), if f (x) has this form, then
f

= 0. Conversely, if f (x) =

a
n
x
n
has a
n
= 0 for some n
not divisible by p, then f

has a nonzero coefcient na
n
and so
f

= 0.
3.39 If R is a commutative ring, dene R[[x]] to be the set of all formal power
series over R.
(i) Show that the formulas dening addition and multiplication on
R[x] make sense for R[[x]], and prove that R[[x]] is a commuta-
tive ring under these operations.
Solution. Each ring axiom can be veried for formal power series,
as in Proposition 3.25 (that a formal power series (s
0
. s
1
. . . .) is
a polynomial, i.e., that its coordinates are eventually 0, does not
enter into the proof).
(ii) Prove that R[x] is a subring of R[[x]].
Solution. The set R[x] is a subset of R[[x]], for every polynomial
over R is a formal power series that is eventually 0. The denitions
of addition and multiplication for power series are the same as
for polynomials. Since 1 R[x] and R[x] is closed under the
operations, it is a subring.
96
(iii) Denote a formal power series = (s
0
. s
1
. s
2
. . . . . s
n
. . . . ) by
= s
0
+s
1
x +s
2
x
2
+ .
Prove that if = 1 + x + x
2
+ , then = 1,(1 x) is in
R[[x]].
Solution. We have
1 + x + x
2
+ = 1 + x(1 + x + x
2
+ ).
Hence, if = 1 + x + x
2
+ , then
= 1 + x.
Solving for gives = 1,(1 x).
3.40 If = (s
0
. s
1
. s
2
. . . . . s
n
. . . . ) is a nonzero formal power series, dene
ord() = m, where m is the smallest natural number for which s
m
= 0.
(i) Prove that if R is a domain, then R[[x]] is a domain.
Solution. If = (s
0
. s
1
. . . . ) and = (t
0
. t
1
. . . . ) are nonzero
power series, then each has an order ( = 0 if and only if it has an
order); let ord() = p and ord() = q. Write
= (c
0
. c
1
. . . .).
For any n 0, we have c
n
=

i +j =n
s
i
t
j
. In particular, if n -
p + q, then i - p and s
i
= 0 or j - q and t
j
= 0; it follows
that c
n
= 0 because each summand s
i
t
j
= 0. The same analysis
shows that c
p+q
= s
p
t
q
, for all the other terms are 0. Since R is a
domain, s
p
= 0 and t
q
= 0 imply s
p
t
q
= 0. Therefore,
ord() = ord() +ord().
(ii) Prove that if k is a eld, then a nonzero formal power series
k[[x]] is a unit if and only if ord() = 0; that is, if its constant
term is nonzero.
Solution. Let u = a
0
+a
1
x +a
2
x
2
+ . If u is a unit, then there
is : = b
0
+b
1
x +b
2
x
2
+ with u: = 1. By Exercise 3.39(iii),
ord(u) +ord(:) = ord(1) = 0.
Since ord() 0 for all (nonzero) k[[x]], it follows that
ord(u) = 0 = ord(:). Therefore, a
0
= 0.
We show that u = a
0
+a
1
x +a
2
x
2
+ is a unit by constructing
the coefcients b
n
of its inverse : = b
0
+ b
1
x + b
2
x
2
+ by
97
induction on n 0. Dene b
0
= a
1
0
. If : exists, then the equa-
tion u: = 1 would imply that

i +j =n
a
i
b
j
= 0 for all n > 0.
Assuming that b
0
. . . . . b
n1
have been dened, then we have
0 = a
0
b
n
+

i +j = n
j - n
a
i
b
j
.
and this can be solved for b
n
because a
0
is invertible.
(iii) Prove that if k[[x]] and ord() = n, then
= x
n
u.
where u is a unit in k[[x]].
Solution. Since ord() = n, we have
= a
n
x
n
+a
n+1
x
n+1
+a
n+2
x
n+2
+
= x
n
(a
n
+a
n+1
x +a
n+2
x
2
+ ).
As a
n
= 0, we have a
n
+a
n+1
x +a
n+2
x
2
+ a unit, by part (ii).
3.41 True or false with reasons.
(i) If R and S are commutative rings and f : R S is a ring ho-
momorphism, then f is also a homomorphism from the additive
group of R to the additive group of S.
Solution. True.
(ii) If R and S are commutative rings and if f is a homomorphism
fromthe additive group of R to the additive group of S with f (1) =
1, then f is a ring homomorphism.
Solution. False.
(iii) If R and S are isomorphic commutative rings, then any ring ho-
momorphism f : R S is an isomorphism.
Solution. False.
(iv) If f : R S is a ring homomorphism, where S is a nonzero ring,
then ker f is a proper ideal in R.
Solution. True.
(v) If I and J are ideals in a commutative ring R, then I J and I J
are also ideals in R.
Solution. False.
(vi) If : R S is a ring homomorphism and if I is an ideal in R,
then (I ) is an ideal in S.
Solution. False.
98
(vii) If : R S is a ring homomorphism and if J is an ideal in S,
then the inverse image
1
(J) is an ideal in R.
Solution. True.
(viii) If R and S are commutative rings, then the projection (r. s) r
is a ring homomorphism R S R.
Solution. True.
(ix) If k is a eld and f : k R is a surjective ring homomorphism,
then R is a eld.
Solution. True.
(x) If f (x) = e
x
, then f is a unit in F(R).
Solution. True.
3.42 Let A be a commutative ring. Prove that a subset J of A is an ideal if and
only if 0 J, u. : J implies u: J, and u J, a A imply au J.
(In order that J be an ideal, u. : J should imply u + : J instead of
u : J.)
Solution. The properties of J differ from those in the denition of an ideal
in that (ii) u, : I implies u : I replaces (ii) u, : I implies
u + : I . Now a = 1, says : J if and only if : J. If (ii) holds,
then u (:) = u + : J, and so (ii) holds. Conversely, if (ii) holds,
then u +(:) = u : J, and so (ii) holds.
3.43 (i) Prove that a eld with 4 elements and I
4
are not isomorphic com-
mutative rings.
Solution. The ring I
4
is not a eld (indeed, it is not even a do-
main), for [2] = [0], but [2] [2] = [0]. But it is easy to check that
any ring isomorphic to a eld must, itself, be a eld.
(ii) Prove that any two elds having exactly four elements are isomor-
phic.
Solution. Let the elements of F be 0, 1, x, and x + 1. Since
we dont yet know that every nite prime eld is isomorphic to
F
p
, for some prime p, we must prove here that 1 + 1 = 0. By
Lagranges theorem, every nonzero element in the additive group
of F has order 2 or 4. In particular, (1 + 1)(1 + 1) = 4 = 0;
were 1 + 1 = 0, then F being a eld, hence a domain, would
be contradicted. It follows that 2x = 0 = 2(x + 1) and that the
additive group of F is isomorphic to the four-group.
The nonzero elements, F

= F {0}, form a cyclic group of

order 3 under multiplication with generator x (of course, x + 1 is
also a generator). Since x
2
must be the element of F

other than
1, we have x
2
= x +1. Similarly, (x +1)
2
= x, for x +1 is also
99
a generator of the cyclic group F

. It is now straightforward to
show that F has the addition and multiplication tables displayed in
the solution to Exercise 3.19, and this is enough to show that any
two elds with four elements are isomorphic, for being isomorphic
means having the same multiplication and addition tables.
3.44 (i) Let : A R be an isomorphism, and let : R A be its
inverse. Show that is an isomorphism.
Solution. Let r. s R. Since is surjective, there are elements
a. b A with (a) = r and (b) = s; since is injective, these
elements a and b are unique. Now an isomorphism gives a +b
the unique element of A with (a +b) = r +s and ab the unique
element of A with (ab) = rs. Therefore, (r + s) = a + b =
(r) + (s) and (rs) = ab = (r)(b). Finally, since 1 is
the unique element of A with (1) = 1, we have (1) = 1. We
conclude that is a ring isomorphism.
(ii) Show that the composite of two homomorphisms (or two isomor-
phisms) is again a homomorphism (or an isomorphism).
Solution. Let
A

B

C
be ring homomorphisms. Now
: 1 (1) = 1 (1) = 1.
so that the composite preserves 1.
Let a. a

A. Then
: a +a

(a +a

)
= (a) +(a

)
(:ar phi (a) +(a

))
= (:ar phi (a)) +((a

)).
One shows that the composite preserves multiplication in the same
way, and one concludes that the composite of ring homomorphisms
is again a ring homomorphism. Since the composite of bijections
is always a bijection, it follows that the composite of ring isomor-
phisms is again an isomorphism.
(iii) Show that A

= R denes an equivalence relation on any family of
commutative rings.
Solution.
(i) Reexive: If A is a ring, then the identity map 1
A
: A A is
an isomorphism.
100
(ii) Symmetric: If : A R is an isomorphism, then part (i)
shows that its inverse is an isomorphism R A.
(iii) Transitive: If
A

B

C
be ring homomorphisms, then part (ii) shows that their com-
posite is an isomorphism A C.
3.45 Let R be a commutative ring and let F(R) be the commutative ring of all
functions f : R R.
(i) Show that R is isomorphic to the subring of F(R) consisting of
all the constant functions.
Solution. If r R, let
r
denote the constant function R R
sending a r for all a R. It is routine to check that : R
F(R), given by r
r
, is an injective homomorphism with the
desired image.
(ii) If f (x) = a
0
+a
1
x+ +a
n
x
n
R[x], let f
.
: R R be dened
by f
.
(r) = a
0
+ a
1
r + + a
n
r
n
[thus, f
.
is the polynomial
function associated to f (x)]. Show that the function : R[x]
F(R), dened by : f (x) f
.
, is a ring homomorphism.
Solution. Absent.
(iii) Show that if R = F
p
, where p is a prime, then x
p
x ker .
Solution. Absent.
3.46 Let R be a commutative ring. Show that the function : R[x] R, de-
ned by
: a
0
+a
1
x +a
2
x
2
+ +a
n
x
n
a
0
.
is a homomorphism. Describe ker in terms of roots of polynomials.
Solution. First of all, (1) = 1. Next, if f (x) =

a
i
x
i
and g(x) =

b
i
x
i
, then f (x) + g(x) =

(a
i
+b
i
)x
i
, and so
( f + g) = a
0
+b
0
= ( f ) +(g).
Finally, since the constant term of f (x)g(x) is a
0
b
0
, we have
( f g) = a
0
b
0
= ( f )(g).
Therefore, is a ring homomorphism.
The kernel of consists of all polynomials having constant term 0; these
are precisely all the polynomials having 0 as a root.
101
3.47 Let : R S be a homomorphism, where R and S are commutative
rings, and let ker = I . If J is an ideal in S, prove that
1
(J) is an ideal
in R which contains I .
Solution. Absent.
3.48 If R is a commutative ring and c R, prove that the function : R[x]
R[x], dened by f (x) f (x + c), is an isomorphism. In more detail,
(

i
s
i
x
i
) =

i
s
i
(x +c)
i
.
Solution. It is clear that (1) = 1. If f (x) =

s
j
x
j
and g(x) =

r
j
x
j
,
then
( f + g) = (

(s
i
+r
i
)x
i
)
=

(s
i
+r
i
)(x +c)
i
=

s
i
(x +c)
i
+

r
i
(x +c)
i
= ( f ) +(g).
as desired.
To prove that ( f g) = (f )(g), rst consider the special case f (x) =
sx
k
, where s R and k 0. If g(x) =

r
i
x
i
, then
(sx
k
f (x)) = (

sr
i
x
i +k
)
=

sr
i
(x +c)
i +k
= s(x +c)
k

r
i
(x +c)
i
= (sx
k
)(g).
We now prove that ( f g) = ( f )(g) by induction on deg( f ). The base
step deg( f ) = 0 is the special case just proved when k = 0. For the
inductive step, write f (x) = h(x)+sx
n+1
, where h(x) = 0 or deg(h) n.
The case h = 0 is the special case above with k = n + 1, and so we may
assume that deg(h) n. Now
( f g) = (hg +sx
n+1
g)
= (hg) +(sx
n+1
g(x)). is additive
= (h)(g) +(sx
n+1
)(g).
induction and the special case
= ((h) +(sx
n+1
))(g)
= (h +sx
n+1
)(g)
= ( f )(g).
102
Therefore, is a ring homomorphism. Finally, is an isomorphism, for
it is easy to see that its inverse is the function given by (

i
s
i
x
i
) =

i
s
i
(x c)
i
.
3.49 If R is a eld, show that R

= Frac(R). More precisely, show that the
homomorphism f : R Frac(R), given by r [r. 1], is an isomorphism.
Solution. If R is a eld, dene : Frac(R) R by [r. s] rs
1
. Note
that rs
1
is dened, because s = 0 and R is a eld. Also, is single
valued, for if [r. s] = [a. b], then rb = sa, hence rs
1
= ab
1
, and so
([r. s]) = ([a. b]). It is routine to check that is a ring homomorphism
inverse to f : R Frac(R).
An alternative solution can be based on the observation that the homo-
morphism f is surjective, for [r. s] = [rs
1
. 1] im f .
3.50 Let R be a domain and let F be a eld containing R as a subring.
(i) Prove that E = {u:
1
: u. : R and : = 0} is a subeld of F
containing R as a subring.
Solution. Absent.
(ii) Prove that Frac(R)

= E, where E is the subeld of F dened in
part (i).
Solution. Absent.
3.51 (i) If A and R are domains and : A R is a ring isomorphism,
then [a. b] [(a). (b)] is a ring isomorphism Frac(A)
Frac(R).
Solution. Each element of Frac(A) has the form [a. b], where
a. b A and b = 0. If : A R is an isomorphism, dene
+: Frac(A) Frac(R) by [a. b] [(a). (b)]. Note that
injective gives (b) = 0, so that [(a). (b)] makes sense.
One must rst check that + is single-valued: if [a. b] = [c. d],
then ad = bc, hence (ab) = (cd), (a)(b) = (c)(d), and
[(a). (b)] = [(c). (d)]. It is straightforward to check that +
is a homomorphism of rings; it is an isomorphism, for +
1
can be
dened explicitly, in a similar fashion.
(ii) Show that a eld k containing an isomorphic copy of Z as a sub-
ring must contain an isomorphic copy of Q.
Solution. The argument just given, except for the last phrase,
shows that if A R are domains, then Frac(A) Frac(R)
[more precisely, there is an injective homomorphism Frac(A)
Frac(R)]. In particular, if Z k, then Q = Frac(Z) Frac(k) = k.
3.52 Let R be a domain with fraction eld F = Frac(R).
(i) Prove that Frac(R[x])

= F(x).
103
Solution. By Exercise 3.49, we may assume that there is an injec-
tive ring homomorphism +: Frac(R[x]) Frac(F(x)) = F(x)
with [ f (x). g(x)] f (x)g(x)
1
. It sufces to show that + is a
surjection. If (x) F[x], then clearing denominators of coef-
cients gives a factorization (x) = r
1
f (x), where r R and
f (x) R[x]. Therefore, if (x)(x)
1
F(x), then there are
r. s R and f (x). g(x) R[x] with
(x)(x)
1
= s f (x)[rg(x)]
1
= +([s f (x). rg(x)]).
(ii) Prove that Frac(R[x
1
. x
2
. . . . . x
n
])

= F(x
1
. x
2
. . . . . x
n
).
Solution. The proof is by induction on n 1, the base step being
part (i). For the inductive step, set S = R[x
1
. x
2
. . . . . x
n
] and the
base step gives
Frac(R[x
1
. x
2
. . . . . x
n+1
]) = Frac(S[x
n+1
])
= Frac(S)(x
n+1
)
= F(x
1
. x
2
. . . . . x
n
)(x
n+1
)
= F(x
1
. x
2
. . . . . x
n
. x
n+1
).
3.53 (i) If R and S are commutative rings, show that their direct product
R S is also a commutative ring.
Solution. Straightforward.
(ii) Show that R {0} is an ideal in R S.
Solution. Either do directly, checking each part of the denition
of ideal, or note that R {0} is the kernel of the homomorphism
R S S given by (r. s) s.
(iii) Show that RS is not a domain if neither R nor S is the zero ring.
Solution. (1. 0) (0. 1) = (0. 0).
3.54 (i) If R and S are commutative rings, prove that
U(R S) = U(R) U(S).
where U(R) is the group of units of R.
Solution. We rst prove that (r. s) is a unit in R S if and only if
r is a unit in R and s is a unit in S.
If r. s) is a unit in R S, then there is (a. b) R S with
(r. s)(a. b) = (1. 1). It follows that ra = 1 and sb = 1; that
is, r is a unit in R and s is a unit in S.
104
Conversely, if r and s are units, then there are a R and b S
with ra = 1 and sb = 1. Therefore, (r. s)(a. b) = (1. 1), and so
(r. s) is a unit in R S.
It follows easily that (r. s) U(R S) if and only if (r. s)
U(R) U(S).
(ii) Show that if m and n are relatively prime, then I
mn

= I
m
I
n
as
rings.
Solution. The isomorphism : I
mn
I
m
I
n
of additive groups
in Theorem 2.128, given by [a] ([a]
m
. [a]
n
) [where [a]
m
de-
notes the congruence class of a mod m],is easily seen to preserve
multiplication: [ab]
m
= [a]
m
[b]
m
and [ab]
n
= [a]
n
[b]
n
.
(iii) Use part (ii) to give a new proof of Corollary 3.54: if (m. n) = 1,
then (mn) = (m)(n), where is the Euler -function.
Solution. If (m. n) = 1, then (U(I
mn
)

= U(I
m
I
n
), by part (i),
and so
U(I
m
I
n
) = U(I
m
) U(I
n
).
But (m) = |U(I
m
)|. and so
(mn) = |U(I
mn
)| = |U(I
m
) U(I
n
)|
= |U(I
m
)||U(I
n
)| = (m)(n).
3.55 (i) Prove that the set F of all 2 2 real matrices of the form A =
_
a b
b a
_
is a eld with operations matrix addition and matrix mul-
tiplication.
Solution. It is easy to check that F is a commutative subring of
the (noncommutative) ring of all 2 2 real matrices (note that the
identity matrix I F). If A = 0, then det(A) = a
2
+b
2
= 0, and
so A
1
exists; since A
1
has the correct form, it lies in F, and so
F is a eld.
(ii) Prove that F is isomorphic to C.
Solution. It is straightforward to check that is a homomor-
phismof elds; it is an isomorphismbecause its inverse is given by
a +i b A.
3.56 True or false with reasons.
(i) If a(x). b(x) F
5
[x] with b(x) = 0, then there exist c(x). d(x)
F
5
[x] with a(x) = b(x)c(x) + d(x), where either d(x) = 0 or
deg(d) - deg(b).
Solution. True.
105
(ii) If g(x). f (x) Z[x] with f (x) = 0, then there exist q(x). r(x)
Z[x] with g(x) = f (x)q(x) + r(x), where either r(x) = 0 or
deg(r) - deg( f ).
Solution. False.
(iii) The gcd of 2x
2
+4x +2 and 4x
2
+12x +8 in Q[x] is 2x +2.
Solution. False.
(iv) If R is a domain, then every unit in R[x] has degree 0.
Solution. True.
(v) If k is a eld and p(x) k[x] is a nonconstant polynomial having
no roots in k, then p(x) is irreducible in k[x].
Solution. False.
(vi) For every quadratic s(x) C[x], there are a. b C and q(x)
C[x] with (x +1)
1000
= s(x)q(x) +ax +b.
Solution. True.
(vii) If k = F
p
(x), where p is a prime, and if f (x). g(x) k[x] satisfy
f (a) = g(a) for all a k, then f (x) = g(x).
Solution. True.
(viii) If k is a eld, then k[x] is a PID.
Solution. True.
(ix) Z is a Euclidean ring.
Solution. True.
(x) There is an integer m such that m
2
1 mod 89.
Solution. True.
3.57 Given a commutative ring R, we saw, in Exercise 3.10, that F(R) =
{all functions R R} is a commutative ring under pointwise operations.
(i) If R is a commutative ring, prove that : R[x] F(R), given by
: f (x) f
.
, is a homomorphism.
Solution. Absent.
(ii) If k is an innite eld, prove that is an injection.
Solution. Absent.
3.58 Find the gcd of x
2
x 2 and x
3
7x + 6 in F
5
[x], and express it as a
linear combination of them.
Solution. The Euclidean algorithm shows that
gcd(x
2
x 2. x
3
7x +6) = x 2.
and
x 2 =
1
4
(x
3
7x +6) +
1
4
(x +1)(x
2
x 2).
106
3.59 Let k be a eld, let f (x) k[x] be nonzero, and let a
1
. a
2
. . . . . a
t
be some
distinct roots of f (x) in k. Prove that
f (x) = (x a
1
)(x a
2
) (x a
t
)g(x)
for some g(x) k[x].
Solution. Absent.
3.60 If R is a domain and f (x) R[x] has degree n, show that f (x) has at most
n roots in R.
Solution. Let F = Frac(R); if f (x) had more than n roots in R, then
it would have more than n roots in F (which contains R), and this is a
contradiction.
3.61 Let R be an arbitrary commutative ring. If f (x) R[x] and if a R
is a root of f (x), that is, f (a) = 0, prove that there is a factorization
f (x) = (x a)g(x) in R[x].
Solution. Absent.
3.62 (i) Show that the following pseudocode implements the Euclidean al-
gorithm nding gcd f (x) and g(x) in k[x], where k is a eld.
Input: g, f
Output: d
d := g; s := f
WHILE s = 0 DO
rem := remainder(d. s)
d := s
s := rem
END WHILE
a := leading coefcient of d
d := a
1
d
Solution. Absent.
(ii) Find ( f. g), where f (x) = x
2
+1. g(x) = x
3
+ x +1 I
3
[x].
Solution. Absent.
3.63 Prove the converse of Euclids lemma. Let k be a eld and let f (x) k[x]
be a polynomial of degree 1; if, whenever f (x) divides a product of
two polynomials, it necessarily divides one of the factors, then f (x) is
irreducible.
Solution. The proof is the same as for the corresponding result for integers.
3.64 Let f (x). g(x) R[x], where R is a domain. If the leading coefcient of
f (x) is a unit in R, then the division algorithm gives a quotient q(x) and a
remainder r(x) after dividing g(x) by f (x). Prove that q(x) and r(x) are
uniquely determined by g(x) and f (x).
107
Solution. If F = Frac(R), use the division algorithm in F[x] to compute
the quotient and remainder for dividing g(x) by f (x). If either quotient or
remainder in R[x] were not unique, then they would not be unique in F[x].
3.65 Let k be a eld, and let f (x). g(x) k[x] be relatively prime. If h(x)
k[x], prove that f (x) | h(x) and g(x) | h(x) imply f (x)g(x) | h(x).
Solution. The proof is the same as for Exercise 1.58, mutatis mutandis.
3.66 If k is a eld in which 1 +1 = 0, prove that

1 x
2
, k(x), where k(x)
is the eld of rational functions.
Solution. Suppose, on the contrary, that

1 x
2
= f (x),g(x), where
f (x). g(x) k[x]; we may assume that f (x),g(x) is in lowest terms;
that is, f (x) and g(x) are relatively prime. Cross multiply and square,
obtaining
f (x)
2
= g(x)
2
(1 x
2
) = g(x)
2
(1 x)(1 + x).
Since 1 +1 = 0, the polynomials 1 x and 1 +x are relatively prime and
irreducible, Euclids lemma gives 1 x | f (x) and 1 + x | f (x); that is,
f (x) = (1 x
2
)h(x)
for some h(x) k[x]. After substituting and canceling,
h(x)
2
(1 x
2
) = g(x)
2
.
Repeat the argument to obtain 1 x | g(x), and this contradicts f (x) and
g(x) being relatively prime. Therefore,

1 x
2
, k(x).
3.67 Let f (x) = (x a
1
) (x a
n
) R[x], where R is a commutative ring.
Show that f (x) has no repeated roots (that is, all the a
i
are distinct) if and
only if the gcd ( f. f

) = 1, where f

is the derivative of f .
Solution. Exercise 3.37 says that x a is a common divisor of f (x) and
f

if and only if (x a)
2
| f (x); that is, if and only if f (x) has repeated
roots.
3.68 Let be the degree function of a Euclidean ring R. If m. n N and m 1,
prove that

is also a degree function on R, where

(x) = m(x) +n for

all x R. Conclude that a Euclidean ring may have no elements of degree
0 or degree 1.
Solution. If x. y R are nonzero, then

(x) = m(x) +n
m(xy) +n
=

(xy).
108
If x. y R and x = 0, there are q. r R with
y = qx +r.
where either r = 0 or (r) - (x). In the second case, it follows that

(r) -

(x), for m 1 > 0. Hence,

is also a degree function on R.

3.69 Let R be a Euclidean ring with degree function .
(i) Prove that (1) (a) for all nonzero a R.
Solution. By the rst axiom in the denition of degree,
(1) (1 r) = (r).
(ii) Prove that a nonzero u R is a unit if and only if (u) = (1).
Solution. If u is a unit, then there is : R with u: = 1; therefore,
(u) (u:) = (1). By part (i), (u) = (1).
Conversely, assume that (u) = (1). By the division algorithm
in R, there are q. r R with 1 = qu +r, where either r = 0 or
(r) - (u). But (u) = (1) and, by part (i), the possibility
(r) - (u) cannot occur. Therefore, r = 0 and u is a unit, for
1 = qu.
3.70 Let =
1
2
_
1 +

19
_
, and let R = Z[].
(i) Prove that N : R

N, dened by N(m+n) = m
2
mn+5n
2
,
is multiplicative: N(u:) = N(u)N(:).
Solution. Absent.
(ii) Prove that the only units in R are 1.
Solution. Absent.
(iii) Prove that there is no surjective ring homomorphism R I
2
or
R I
3
.
Solution. Absent.
(iv) Assume that R has a degree function : R

N. Choose u
R {0. 1. 1} with (u) minimal. Prove, for all r R that there
exists d {0. 1. 1} such that r d (u).
Solution. Absent.
(v) Use the ring R,(u) to prove that R is not a Euclidean ring.
Solution. Absent.
3.71 Let R be a Euclidean ring with degree function , and assume that b R
is neither zero nor a unit. Prove, for every i 0, that (b
i
) - (b
i +1
).
109
Solution. There are q. r R with b
i
= qb
i +1
+r, where either r = 0 or
(r) - (b
i +1
). If r = 0, then b
i
= qb
i +1
; canceling, 1 = qb, and this
contradicts the hypothesis that b is not a unit. Therefore, r = 0. But
r = b
i
qb
i +1
= b
i
(1 qb).
so that
(b
i
) (b
i
(1 qb)) = (r)
(1 qb is nonzero because b is not a unit). Thus,
(b
i
) (r) - (b
i +1
).
3.72 If k is a eld, prove that the ring of formal power series k[[x]] is a PID.
Solution. Let I be an ideal in k[[x]]; we may assume that I = {0} because
{0} = (0). Choose I of smallest order, say, = a
m
x
m
+a
m+1
x
m+1
+
= x
m
:, where : = a
m
+ a
m+1
x + ; note that : is a unit in k[[x]],
by Exercise 3.40. We claim that I = (x
m
). Now x
m
= :
1
I , so that
(x
m
) I . For the reverse inclusion, if k[[x]] is nonzero, then =
b
n
x
n
+b
n+1
x
n+1
+ = x
n
n, where n m and n = b
n
+b
n+1
x + .
Hence, = x
m
x
nm
n (x
m
), so that I (x
n
). Therefore, every ideal in
k[[x]] is principal.
3.73 Let k be a eld, and let polynomials a
1
(x), a
2
(x), . . . , a
n
(x) in k[x] be
given.
(i) Show that the greatest common divisor d(x) of these polynomials
has the form

t
i
(x)a
i
(x), where t
i
(x) k[x] for 1 i n.
Solution. The proof is the same as for Exercise 1.65(i), mutatis
mutandis.
(ii) Prove that if c(x) is a monic common divisor of these polynomials,
then c(x) is a divisor of d(x).
Solution. The proof is the same as for Exercise 1.65(ii), mutatis
mutandis.
3.74 Let [ f (x). g(x)] denote the lcm of f (x). g(x) k[x], where k is a eld.
Show that if f (x)g(x) is monic, then
[ f. g]( f. g) = f g.
Solution. The proof is the same as for Exercise 1.76, mutatis mutandis.
3.75 If k is a eld, show that the ideal (x. y) in k[x. y] is not a principal ideal.
Solution. If the ideal (x. y) in k[x. y] is a principal ideal, then there
is d = d(x. y) that generates it. Thus, x = d(x. y) f (x. y) and y =
d(x. y)g(x. y) for f (x. y). g(x. y) k[x. y]. Taking degrees in each vari-
able, deg
x
(d) 1 and deg
y
(d) 1, and so d(x. y) = ax + by + c, for
110
some constant c. If x is a multiple of ax +by +c, then b = 0; if y is a mul-
tiple, then a = 0. We conclude that d(x. y) is a nonzero constant. Since k
is a eld, d is a unit, and so (x. y) = (d) = k[x. y], a contradiction.
3.76 For every m 1, prove that every ideal in I
m
is a principal ideal. (If m is
composite, then I
m
is not a PID because it is not a domain.)
Solution. Let I be an ideal in I
m
. Consider the ring homomorphism
: Z I
m
dened by : n [n]; of course, is a surjection. It is
easily seen that the inverse image
J =
1
(I ) = {n Z : (n) I }
is an ideal in Z; moreover, (J) = I . But every ideal in Z is principal, so
that J = (a). It follows that I = ([a]).
This argument really shows that if R is a PID and L is an ideal in R, then
every ideal in the quotient ring R,L is principal.
3.77 Let R be a PID and let R be an irreducible element. If R and
, prove that and are relatively prime.
Solution.
3.78 (i) Show that x. y k[x. y] are relatively prime but that 1 is not a lin-
ear combination of them; that is, there do not exist s(x. y). t (x. y)
k[x. y] with 1 = xs(x. y) + yt (x. y).
Solution. As in the solution to Exercise 3.75, a degree argument
shows that the only divisors of x are of the form ax, where a k,
and the only divisors of y are of the form by, where b k. It
follows that the only monic common divisor is 1, and so x and y
are relatively prime. Another degree argument shows that there
are not polynomials s and t with xs(x. y) + yt (x. y) = 1.
(ii) Show that 2 and x are relatively prime in Z[x], but that 1 is not a
linear combination of them; that is, there do not exist s(x). t (x)
Z[x] with 1 = 2s(x) + xt (x).
Solution. The argument is analogous to that in part (i).
3.79 Because x 1 = (

x + 1)(

x 1), a student claims that x 1 is not

irreducible. Explain the error of his ways.
Solution. Irreducibility of a polynomial f (x) k[x] is dened in terms of
possible factorizations f (x) = g(x)h(x), where g(x) and h(x) are poly-
nomials. We claim that

x , k[x]; otherwise,

x would have some de-
gree, say n, which is an integer. But the equation

x

x = x gives, by
Lemma 3.24, 2n = 1, a contradiction. It follows that neither

x + 1 nor

x 1 is a polynomial.
3.80 (i) Factor each of the integers 5 and 13 as a product of two nonunits
in Z[x], and factor 65 as a product of four nonunits in Z[x].
111
Solution. Absent.
(ii) Find two different ways of grouping the factors of 65 in part (i)
as a product of two conjugate factors. Use these to get two
different expressions for 65 as a sum of two squares in Z.
Solution. Absent.
3.81 Prove that there are domains R containing a pair of elements having no
gcd.
Solution. Let k be a eld, and let R be the subring of k[x] consisting of
all polynomials having no linear term; that is, f (x) R if and only if
f (x) = s
0
+s
2
x
2
+s
3
x
3
+ . We claim that x
5
and x
6
have no gcd: their
only monic divisors are 1. x
2
. and x
3
, none of which is divisible in R by
the other two. For example, x
2
is not a divisor of x
3
, for if x
3
= f (x)x
2
,
then (in k[x]) we have deg( f ) = 1. But there are no linear polynomials
in R.
3.82 True or false with reasons.
(i) Every element of Z[x] is a product of a constant in Z and monic
irreducible polynomials in Z[x].
Solution. False.
(ii) Every element of Z[x] is a product of a constant in Z and monic
irreducible polynomials in Q[x].
Solution. True.
(iii) If k is a eld and f (x) k[x] can be written as ap
1
(x) p
m
(x)
and bq
1
(x) q
n
(x), where a. b are constants, p
1
(x). . . . . p
m
(x)
are monic irreducible polynomials, and q
1
(x). . . . . q
n
(x) are monic
nonconstant polynomials, then q
1
(x). . . . . q
n
(x) are irreducible.
Solution. False.
(iv) If k is a eld and f (x) k[x] can be written as ap
1
(x) p
m
(x)
and bq
1
(x) q
n
(x), where a. b are constants, p
1
(x). . . . . p
m
(x)
are monic irreducible polynomials, and q
1
(x). . . . . q
n
(x) are monic
nonconstant polynomials, then m n.
Solution. True.
(v) If k is a subeld of K and f (x) k[x] has the factorization
f (x) = ap
e
1
1
p
e
n
n
, where a is a constant and the p
i
(x) are
monic irreducible in k[x], then f (x) = ap
e
1
1
p
e
n
n
is also the
factorization of f (x) in K[x] as a product of a constant and monic
irreducible polynomials.
Solution. False.
112
(vi) If f (x) is a polynomial over a eld K whose factorization into
a constant and monic irreducible polynomials in K[x] is f (x) =
ap
e
1
1
p
e
n
n
, and if all the coefcients of f (x) and of the polyno-
mials p
i
(x) lie in some subeld k K, then f (x) = ap
e
1
1
p
e
n
n
is also the factorization of f (x) in k[x] as a product of a constant
and monic irreducible polynomials.
Solution. True.
3.83 In k[x], where k is a eld, let g = p
e
1
1
p
e
m
m
and h = p
f
1
1
p
f
m
m
, where
the p
i
s are distinct monic irreducibles and e
i
, f
i
0 for all i . Prove that
g | h if and only if e
i
f
i
for all i .
Solution. The proof for polynomials is essentially the same as for integers.
3.84 (i) If f (x) R[x], show that f (x) has no repeated roots in C if and
only if ( f. f

) = 1.
Solution. By part (ii), f (x) has no repeated roots in C[x] if and
only if ( f. f

) = 1 in C[x]. By Corollary 3.75, one can check this
criterion by computing this gcd in R[x].
(ii) Prove that if p(x) Q[x] is an irreducible polynomial, then p(x)
has no repeated roots.
Solution. If p(x) =

n
i =0
a
i
x
i
, then p

(x) =

n
i =1
i a
i
x
i 1
. It
follows that deg( p

) = n 1, and so p

is not the zero polynomial.

Since p(x) is irreducible, the gcd ( p. p

) = 1, and so part (ii)

applies to show that p(x) has no repeated roots.
Note that this argument may fail if Q is replaced by F
p
, where p
is prime.
3.85 Let = e
2i ,n
.
(i) Prove that
x
n
1 = (x 1)(x )(x
2
) (x
n1
)
and, if n is odd, that
x
n
+1 = (x +1)(x + )(x +
2
) (x +
n1
).
Solution. The n numbers 1. .
2
. . . . .
n1
are all distinct. But
they are all roots of x
n
1, and so Theorem 3.50 gives the rst
equation:
x
n
1 = (x 1)(x )(x
2
) (x
n1
).
If n is odd, then replace x by x to get
(x)
n
1 = (x 1)(x )(x
2
) (x
n1
)
= (1)
n
(x +1)(x + )(x +
2
) (x +
n1
).
113
Since n is odd,
(x)
n
1 = x
n
1 = (x
n
+1).
and one can now cancel the minus sign from each side.
(ii) For numbers a and b, prove that
a
n
b
n
= (a b)(a b)(a
2
b) (a
n1
b)
and, if n is odd, that
a
n
+b
n
= (a +b)(a + b)(a +
2
b) (a +
n1
b).
Solution. If b = 0, then both sides equal a
n
; if b = 0, then set
x = a,b in part (ii).
3.86 True or false with reasons.
(i)

3 is an algebraic integer.
Solution. True.
(ii)
13
78
is not a rational root of 1 +5x +6x
2
.
Solution. False.
(iii) If f (x) = 3x
4
+ ax
3
+ bx
2
+ cx + 7 with a. b. c Z, then the
roots of f (x) in Q, if any, lie in {1. 7.
1
3
.
7
3
}.
Solution. True.
(iv) If f (x) = 3x
4
+ ax
3
+ bx
2
+ cx + 7 with a. b. c Q, then the
roots of f (x) in Q, if any, lie in {1. 7.
1
3
.
7
3
}.
Solution. False.
(v) 6x
2
+10x +15 is a primitive polynomial.
Solution. True.
(vi) Every primitive polynomial in Z[x] is irreducible.
Solution. False.
(vii) Every irreducible polynomial in Z[x] is primitive.
Solution. False.
(viii) Every monic polynomial in Z[x] is primitive.
Solution. True.
(ix) The content of 3x +
1
5
is
3
5
.
Solution. False.
(x) The content of 3x +
6
5
is
3
5
.
Solution. True.
114
(xi) If f (x) = g(x)h(x) in Q[x], and if f (x) has all its coefcients in
Z, then all the coefcients of g(x) and h(x) also lie in Z.
Solution. False.
(xii) For every integer c, the polynomial (x + c)
2
(x + c) 1 is
irreducible in Q[x].
Solution. True.
(xiii) For all integers n, the polynomial x
8
+5x
3
+5n is irreducible in
Q[x].
Solution. False.
(xiv) The polynomial x
7
+ 9x
3
+ (9n + 6) is irreducible in Q[x] for
every integer n.
Solution. True.
3.87 Determine whether the following polynomials are irreducible in Q[x].
(i) f (x) = 3x
2
7x 5.
Solution. There are no rational roots: the candidates are
1. 5.
1
3
.
5
3
.
Therefore, f (x) is irreducible, by Proposition 3.65.
(ii) f (x) = 350x
3
25x
2
+34x +1.
Solution. Absent.
(iii) f (x) = 2x
3
x 6.
Solution. There are no rational roots: the candidates are

1
2
. 1.
3
2
. 2. 3. 6.
Therefore, f (x) is irreducible, by Proposition 3.65.
(iv) f (x) = 8x
3
6x 1.
Solution. f (x) x
3
+ x 1 mod 7. None of the elements of
F
7
is a root, so that f (x) is irreducible in F
7
[x] (because a cubic
is irreducible if it has no roots, by Proposition 3.65). Therefore,
f (x) is irreducible over Q.
(v) f (x) = x
3
+6x
2
+5x +25.
Solution. f (x) x
3
+x +1 mod 2. Since there are no roots over
F
2
, the cubic f (x) is irreducible over Q, by Proposition 3.65.
(vi) f (x) = x
5
4x +2.
Solution. f (x) is irreducible, by the Eisenstein criterion with
p = 2.
115
(vii) f (x) = x
4
+ x
2
+ x +1.
Solution. First, f (x) has no rational roots, by Theorem 3.90 (the
only candidates are 1), and so we must show it is not a product of
two (irreducible) quadratics. If f (x) factors into quadratics, then
x
4
10x
2
+1 = (x
2
+ax +b)(x
2
ax +c).
where a. b. c Q (the coefcient of x in the second factor must
be a because f (x) has no cubic term). Expanding and equating
coefcients gives
b +c a
2
= 10.
a(c b) = 0.
bc = 1.
Now b = 1,c, so that ac = ab = a,c, and
ac
2
= a.
Hence, either a = 0 or c
2
= 1. If a = 0, then we have
b +c = 10 and bc = 1.
These equations lead to a quadratic, c
2
10c + 1, whose roots
are irrational, contradicting c Q. Therefore, a = 0 and b = c.
Since bc = 1, this gives b = c = 1 or b = c = 1. Now
c +b a
2
= 10, so there are only two possibilities:
2 a
2
= 10 or 2 a
2
= 10.
Hence, either a
2
= 12 or a
2
= 8, each of which forces a to be
irrational. Therefore, there is no such factorization, and f (x) is
irreducible over Q.
(viii) f (x) = x
4
10x
2
+1.
Solution. f (x) has no rational roots, for the only candidates are
1. Suppose that
x
4
10x
2
+1 = (x
2
+ax +b)(x
2
ax +c) in Q[x]
(we may assume the coefcient of x in the second factor is a
because f (x) has no cubic term). Expanding and equating coef-
cients gives the following equations:
c +b a
2
= 10
a(c b) = 0
bc = 1.
116
The middle equation gives a(c b) = 0, so that either a = 0 or
b = c. In the rst case, we obtain
c +b = 10
cb = 1.
Substituting c = b
1
, the rst equation gives b
2
10b + 1 = 0.
But the quadratic formula gives b = 5 2

6, which is irrational.
On the other hand,if b = c, then bc = 1 implies b = 1 = c. The
rst equation gives a
2
= 102 - 0, and this is also impossible.
We conclude that there is no factorization of f (x) in Q[x].
(ix) f (x) = x
6
210x 616.
Solution. Eisensteins criterion applies, for 7 | 210 and 7 | 616,
but 7
2
616.
(x) f (x) = 350x
3
+ x
2
+4x +1.
Solution. Reducing mod 3 to gives an irreducible cubic in F
3
[x].
3.88 If p is a prime, prove that there are exactly
1
3
_
p
3
p
_
monic irreducible
cubic polynomials in F
p
[x].
Solution. Absent.
3.89 Prove that there are exactly 6 irreducible quintics in F
2
[x].
Solution. There are 32 quintics in F
2
[x], 16 of which have constant term 0;
that is, have 0 as a root. Of the 16 remaining polynomials, we may discard
those having an even number of nonzero terms, for 1 is a root of these; and
now there are 8. If a quintic f (x) with no roots is not irreducible, then its
factors are irreducible polynomials of degrees 2 and 3; that is,
f (x) = (x
2
+ x +1)(x
3
+ x +1) = x
5
+ x
4
+1.
or
f (x) = (x
2
+ x +1)(x
3
+ x
2
+1) = x
5
+ x +1.
Thus, the irreducible polynomials are:
x
5
+ x
3
+ x
2
+ x +1 x
5
+ x
4
+ x
2
+ x +1
x
5
+ x
4
+ x
3
+ x +1 x
5
+ x
4
+ x
3
+ x
2
+1
x
5
+ x
3
+1 x
5
+ x
2
+1.
3.90 (i) If a = 1 is a squarefree integer, show that x
n
a is irreducible
in Q[x] for every n 1. Conclude that there are irreducible poly-
nomials in Q[x] of every degree n 1.
117
Solution. Since a = 0 and a = 1, there is a prime p dividing
a; since a is squarefree, p
2
a. As all coefcients of x
i
for 0 -
i - n are 0, Eisensteins criterion applies to show that x
n
a is
irreducible in Q[x].
(ii) If a = 1 is a squarefree integer, prove that
n

a is irrational.
Solution. Since x
n
a is irreducible in Q[x], it has no rational
roots.
3.91 Let k be a eld, and let f (x) = a
0
+a
1
x + +a
n
x
n
k[x] have degree
n. If f (x) is irreducible, then so is a
n
+a
n1
x + +a
0
x
n
.
Solution. If f

(x) denotes the polynomial f (x) with coefcients reversed,
then a factorization f

(x) = g(x)h(x) gives a factorization f (x). One
sees this just by using the denition of multiplication of polynomials. Let
g(x) =

p
i =0
b
i
x
i
and h(x) =

q
j =0
c
j
x
j
, where p +q = n. Thus,
a
nm
=

i +j = m
b
i
c
j
.
It follows that

i +j = nm
b
pi
c
qj
= a
n(nm)
= a
k
.
Therefore, if we dene g

(x) =

p
i =0
b
pi
x
i
and h

(x) =

q
j =0
c
qj
x
j
,
then f (x) = g

(x)h

(x), contradicting the irreducibility of f (x).

Note that f (x) f

(x), which reverses coefcients, is not a well-
dened function k[x] k[x], because it is not clear how to dene f

(x)
if the constant term of f (x) is zero. And even if one makes a bona de
denition, the function is not a homomorphism. For example, let f (x) =
x
5
+3x
4
; that is, in sequence notation,
f (x) = (0. 0. 0. 0. 3. 1. 0. . . .).
Let g(x) = x
3
+ x; in sequence notation,
g(x) = (0. 1. 0. 1. 0. . . .).
Now f (x)g(x) = [x
8
+3x
7
+ x
6
+4x
5
+3x
4
; in sequence notation,
f (x)g(x) = (0. 0. 0. 0. 3. 4. 1. 3. 1. 0. . . .).
Therefore,
[ f (x)g(x)]

= 3x
4
+4x
3
+ x
2
+3x +1.
which is a quartic. But f

(x) = 3x + 1 and g

(x) = x
2
+ 1, so that
f

(x)g

(x) is a cubic. Therefore, [ f g]

= f

g

.
118
3.92 True or false with reasons.
(i) If I is a proper ideal in a commutative ring R and : R R,I
is the natural map, then ker = I .
Solution. True.
(ii) If I is a proper ideal in a commutative ring R and : R R,I
is the natural map, then is surjective.
Solution. True.
(iii) If f : R S is a homomorphism of commutative rings, then S
has a subring isomorphic to R,(ker f ).
Solution. True.
(iv) If I is a proper ideal in a commutative ring R, then R has a subring
isomorphic to R,I .
Solution. False.
(v) If p is a prime number, then every eld of characteristic p is nite.
Solution. False.
(vi) Every eld of characteristic 0 is innite.
Solution. True.
(vii) If f (x) is an irreducible polynomial over a eld k, then k[x],( f (x))
is a eld.
Solution. True.
(viii) If f (x) is a nonconstant polynomial over a eld k and if the quo-
tient ring k[x],( f (x)) is a eld, then f (x) is irreducible.
Solution. True.
(ix) If f (x) is an irreducible polynomial over a eld k, then every ele-
ment z k[x],( f (x)) is a root of f (x).
Solution. False.
(x) If k K are elds and z K is a root of some nonzero polyno-
mial p(x) k[x], then p(x) is irreducible in k[x].
Solution. False.
(xi) There is a eld containing C(x) and

x +i .
Solution. True.
(xii) For every positive integer n, there exists a eld with exactly 11
n
elements.
Solution. True.
119
(xiii) For every positive integer n, there exists a eld with exactly 10
n
elements.
Solution. False.
(xiv) For every positive integer n, there exists a eld with exactly 9
n
elements.
Solution. True.
(xv) There exists a eld E of characteristic 2 such that x
4
+ x +1 is a
product of linear factors in E[x].
Solution. True.
3.93 For every commutative ring R, prove that R[x],(x)

= R.
Solution. Dene : R[x] R by
f (x) constant term of f (x).
It is easy to check that is a surjective homomorphism with ker = (x).
3.94 (Chinese Remainder Theorem in k[x])
(i) Prove that if k is a eld and f (x), f

(x) k[x] are relatively
prime, then given b(x). b

(x) k[x], there exists c(x) k[x]

with
c b ( f ) and c b

( f

);
moreover, if d(x) is another common solution, then cd ( f f

).
Solution. Every solution of c b ( f ) has the form c = b +g f ,
where g(x) k[x]. Hence, we must nd g such that
g f (b

b) ( f

).
Since ( f. f

) = 1, there are polynomials s. t k[x] with s f +
t f

= 1. Dene g = s(b

b). Then
g f = s f (b

b)
= (1 t f

)(b

b)
= (b

b) t (b

b) f

( f

).
If d is another common solution, then both f and f

divide c d.
By Exercise 3.94, f f

| c d, and so c d ( f f

).
(ii) Prove that if k is a eld and f (x), g(x) k[x] are relatively prime,
then
k[x],( f (x)g(x))

= k[x],( f (x)) k[x],(g(x)).
Solution. We adapt the proof of Theorem 2.131.
120
If a(x) k[x], denote its congruence class in k[x],( f (x)) by
[a]
f
. It is easy to check that
: k[x] k[x],( f (x)) k[x],(g(x)).
given by
a ([a]
f
. [a]
g
).
is a ring homomorphism. We claim that ker = ( f g). Clearly,
( f g) ker . For the reverse inclusion, if a ker , then [a]
f
=
[0]
f
and [a]
g
= [0]
g
; that is, a ( f ) and a (g); that is, f | a
and g | a. Since f and g are relatively prime, Exercise 3.94 gives
f g | a, and so a ( f g), that is, ker ( f g) and ker = ( f g).
We now show that is surjective. If
([a]
f
. [b]
g
) k[x],( f (x)) k[x],(g(x)).
is there h(x) k[x] with
(h) = ([h]
f
. [h]
g
) = ([a]
f
. [b]
g
);
that is, is there h k[x] with h a ( f ) and h b (g)? Since
f and g are relatively prime, part (i) provides a solution h. The
rst isomorphism theorem now gives
k[x],( f (x)g(x)) =

= k[x],( f (x)) k[x],(g(x)).

3.95 Generalize Exercise 3.84 by proving that if k is a eld of characteristic 0
and if p(x) k[x] is an irreducible polynomial, then p(x) has no repeated
roots.
Solution. Absent.
3.96 (i) Prove that a eld K cannot have subelds k

and k

with k

= Q
and k

= F
p
for some prime p.
Solution. Suppose such subelds k

and k

exist, and consider

their intersection k = k

. The one element 1 lies in k; since

k k

, we have 0 = p 1 = 1 + 1 + + 1, where there are p

summands equal to 1. On the other hand, in k

, we have p 1 = 0.
This is a contradiction.
(ii) Prove that a eld K cannot have subelds k

and k

with k

= F
p
and k

= F
q
, where p = q.
Solution. The argument in part (i) also works here: if p - q, then
p 1 = 0 in k = k

, because k k

, but p 1 = 0 in k because
k k

.
121
3.97 Let p be a prime and let q = p
n
for some n 1.
(i) Show that the function F : F
q
F
q
, given by F(a) = a
p
, is an
isomorphism.
Solution. We have F(1) = 1
p
= 1 and
F(xy) = (xy)
p
= x
p
y
p
= F(x)F(y).
Furthermore,
F(x + y) = (x + y)
p
= x
p
+ y
p
.
by Proposition 1.63.
Now F is an injection, for if 0 = F(x) = x
p
, then x = 0. Since
F
p
is nite, F is an isomorphism, by Exercise 2.13.
(ii) Showthat every element a F
q
has a pth root, i.e., there is b F
q
with a = b
p
.
Solution. Each a F
q
has a pth root because F is surjective.
(iii) Let k be a eld of characteristic p > 0. For every positive inte-
ger n, show that the ring homomorphism F
n
: k k, given by
F
n
(a) = a
p
n
, is injective.
Solution. The proof is a straightforward induction on n 1, the
base step being part (i).
3.98 Prove that every element z in a nite eld E is a sum of two squares. (If
z = a
2
is a square, then we may write z = a
2
+0
2
.)
Solution. Absent.
3.99 If p is a prime and p 3 mod 4, prove that either a
2
2 mod p is
solvable or a
2
2 mod p is solvable.
Solution. Since p

= 3 mod 4, we have p = 4k +3 for some integer k, and
so
p 1 = (4k +3) 1 = 4k +2 = 2(2k +1);
that is, p 1 = 2m, where m = 2k +1 is odd. By Theorem 2.131, F

p
=
1 H for some subgroup H of order m. Since F
2
I
m
= ({1} H)
({1} H), however, either 2 H or 2 H. But Exercise 2.82 says
that every element in H has a square root. If 2 has a square root, then there
is an integer a with a
2
= 2 in F
p
. and this says that a
2
2 mod p.
3.100 (i) Prove that x
4
+1 factors in F
2
[x].
Solution. x
4
+1 factors in F
2
[x] because 1 is a root.
122
(ii) If x
4
+1 = (x
2
+ax +b)(x
2
+cx +d) F
p
[x], where p is an
odd prime, prove that c = a and
d +b a
2
= 0
a(d b) = 0
bd = 1.
Solution. After expanding, we obtain the equations
a +c = 0
d +ac +b = 0
ad +bc = 0
bd = 1.
Substituting c = a yields the desired equations.
(iii) Prove that x
4
+1 factors in F
p
[x], where p is an odd prime, if any
of the following congruences are solvable:
b
2
1 mod p.
a
2
2 mod p.
Solution. Using the hint, x
4
+ 1 = (x
2
+ b)(x
2
b) in F
p
[x]
if a 0 mod p. If a 0 mod p, then d b mod p and
b
2
bd 1 mod p. Hence, b 1 mod p, by Lemma 3.82
and Exercise 3.100. From part (i), we have a
2
2 mod p. As-
suming that one of these congruences can be solved, we have fac-
torizations in F
p
[x]
x
4
+1 = (x
2
+ax +1)(x
2
ax +1)
or
x
4
+1 = (x
2
+ax 1)(x
2
ax 1).
(iv) Prove that x
4
+1 factors in F
p
[x] for all primes p.
Solution. In part (i), we saw that x
4
+ 1 factors in F
2
[x], and in
part (iii), we saw that it factors in F
p
[x] for every odd prime p if
certain congruences can be solved. As every odd prime (indeed,
every odd integer) is congruent to 1 or 3 mod 4, Exercise 3.100
completes the proof.
123
3.101 Generalize Proposition 3.116(iii) as follows. Let : k k

be an iso-
morphism of elds, let E,k and E

,k

be extensions, let p(x) k[x]

and p

(x) k

[x] be irreducible polynomials (as in Theorem 3.33, if

p(x) =

a
i
x
i
, then p

(x) =

(a
i
)x
i
), and let z E and z

be roots of p(x). p

(x), respectively. Then there exists an isomorphism

: k(z) k

(z

) with (z) = z

and with extending .

k(z)

//
k

(z

)
k

//
k

Solution. Absent.
3.102 Let f (x) = a
0
+ a
1
x + + a
n1
x
n1
+ x
n
k[x], where k is a eld,
and suppose that f (x) = (x r
1
)(x r
2
) . . . (x r
n
) E[x], where E is
some eld containing k. Prove that
a
n1
= (r
1
+r
2
+ +r
n
) and a
0
= (1)
n
r
1
r
2
r
n
.
Conclude that the sum and the product of all the roots of f (x) lie in k.
Solution. Both formulas are proved by induction on n 1.
3.103 If E = F
2
[x],( p(x)), where p(x) = x
3
+ x + 1, then E is a eld with 8
elements. Show that a root of p(x) is a primitive element of E by writing
every nonzero element of E as a power of .
Solution. See Example 4.127.
3.104 (i) Prove, for all n 1, that there is an irreducible polynomial of
degree n in Q[x].
Solution.
(ii) Prove, for all n 1 and every prime p, that there is an irreducible
polynomial of degree n in F
p
[x].
Solution. Use the existence of a eld with exactly p
n
elements.
(iii) Prove, for all n 1 and every nite eld k, that there is an irre-
ducible polynomial of degree n in k[x].
3.105 If E is a nite eld, use Cauchys theorem to prove that |E| = p
n
for some
prime p and some n 1.
Solution. If k is the prime eld of E, then Proposition 3.110 says that
k

= Q or k

= F
p
for some prime p; since Q is innite, we have k of
characteristic p. Therefore, pa = 0 for all a E; that is, as an additive
abelian group, every nonzero element in E has order p. If there is a prime
divisor q of |E| with q = p, then Cauchys theorem gives a nonzero el-
ement b E with qb = 0, contradicting every nonzero element having
order p. We conclude that |E| = p
n
for some n 1.
124
Exercises for Chapter 4
4.1 True or false with reasons.
(i) If k is a eld, then the subset E of all all polynomials of odd degree
is a subspace of k[x].
Solution. True.
(ii) If A and B are n n matrices over a eld k, and if the homoge-
neous system Ax = 0 has a nontrivial solution, then the homoge-
neous system (BA)x = 0 has a nontrivial solution.
Solution. True.
(iii) If A and B are n n matrices over a eld k, and if the homoge-
neous system Ax = 0 has a nontrivial solution, then the homoge-
neous system (AB)x = 0 has a nontrivial solution.
Solution. False.
(iv) If :
1
. :
2
. :
3
. :
4
spans a vector space V, then dim(V) = 4.
Solution. False.
(v) If k is a eld, then the list 1. x. x
2
. . . . . x
100
is linearly indepen-
dent in k[x].
Solution. True.
(vi) There is a linearly independent list of 4 matrices in Mat
2
(R).
Solution. True.
(vii) There is a linearly independent list of 5 matrices in Mat
2
(R).
Solution. False.
(viii) [Q(E
2i ,5
) : Q] = 5.
Solution. False.
(ix) There is an inner product on R
2
with (:. :) = 0 for some nonzero
: R
2
.
Solution. True.
(x) The set of all f : R R with f (1) = 0 is a subspace of F(R).
Solution. True.
4.2 (i) If f : k k is a function, where k is a eld, and if k, dene
a new function f : k k by a f (a). Prove that with this
denition of scalar multiplication, the ring F(k) of all functions
on k is a vector space over k.
Solution. Straightforward verication of the axioms.
125
(ii) If PF(k) F(k) denotes the family of all polynomial functions
a
n
a
n
+ +
1
a +
0
, prove that PF(k) is a subspace of
F(k).
Solution. P(k) is closed under addition and scalar multiplication
4.3 Prove that dim(V) 1 if and only if the only subspaces of a vector space
V are {0} and V itself.
Solution. Assume that dim(V) = 1 (the result is obvious if dim(V) = 0).
Let e be a basis of V, let {0} = S V, and let s = 0 lie in S. The list e. s
must be linearly dependent, so that there are scalars . , not both 0, with
e + s = 0. Now = 0, lest s = 0 and = 0 (for s = 0). Hence,
e =
1
s S, forcing V S, and so S = V.
Conversely, suppose the only subspaces of V are {0} and V. Assume
V = {0} (otherwise dim(V) = 0 and we are done), and choose : V with
: = 0. If u V, then : = V = :. u (for neither subspace is {0}), so
that :. u is a linearly dependent list. Therefore, dim(V) = 1.
4.4 Prove, in the presence of all the other axioms in the denition of vector
space, that the commutative law for vector addition is redundant; that is, if
V satises all the other axioms, then u +: = : +u for all u. : V.
Solution. If u. : V, evaluate [(:) +(u)] in two ways.
4.5 Is L a subspace of Mat
n
(k) if L is the subset consisting of all the n n
Latin squares?
Solution. No.
4.6 (i) If V is a vector space over F
2
and if :
1
= :
2
are nonzero vectors in
V, prove that :
1
. :
2
is linearly independent. Is this true for vector
spaces over any other eld?
Solution. The statement is false for every eld other than F
2
.
(ii) Let k be a eld, and let P
2
(k) be the projective plane consisting of
all points [x] for x k
3
(as in Example 4.26). Prove that [x] = [y]
in P
2
(k) if and only if x. y is a linearly independent list in k
3
.
Solution. Absent.
4.7 Prove that the columns of an m n matrix A over a eld k are linearly
dependent in k
m
if and only if the homogeneous system Ax = 0 has a
nontrivial solution.
Solution. Absent.
4.8 (i) Prove that the list of polynomials 1. x. x
2
. x
3
. . . . . x
100
is a lin-
early independent list in k[x], where k is a eld.
Solution. A linear combination of X = 1. x. x
2
. x
3
. . . . . x
100
is
a polynomial f (x) of degree 100; the coefcients in the linear
combination are precisely the coefcients of f (x). But f (x) = 0
126
if and only if all of its coefcients are 0; therefore, X is linearly
independent.
(ii) Dene V
n
= 1. x. x
2
. . . . . x
n
. Prove that 1. x. x
2
. . . . . x
n
is a
basis of V
n
, and conclude that dim(V
n
) = n +1.
Solution. Absent.
4.9 It is shown in analytic geometry that if
1
and
2
are nonvertical lines with
slopes m
1
and m
2
, respectively, then
1
and
2
are perpendicular if and
only if m
1
m
2
= 1. If

i
= {:
i
+u
i
: R}.
for i = 1. 2. prove that m
1
m
2
= 1 if and only if the dot product :
1
:
2
=
0.
Solution. The lines
1
and
2
are perpendicular if and only if :
1
and :
2
are
perpendicular if and only if :
1
:
2
= 0. If :
1
= (a. b), then m
1
= b,a (the
slope of the line joining the origin O with (a. b)), and if :
2
= (c. d), then
m
2
= d,c. Hence,
m
1
m
2
= 1 1 = (b,a)(d,c)
b,a = c,d
ac = bd
ac +bd = 0
:
1
:
2
= 0

1
and
2
are perpendicular.
4.10 (i) A line in space passing through a point u is dened as
{u +n : R} R
3
.
where n is a xed nonzero vector. Show that every line through u
is a coset of a one-dimensional subspace of R
3
.
Solution. If the origin (0. 0. 0) lies on a line , then u = 0 and
= {n : R}, where n is some xed nonzero vector. In
vector space notation, = n, a one-dimensional subspace of
R
3
. (The converse is true, and it is easy to prove.)
(ii) A plane in space passing through a point u is dened as the subset
{: R
3
: (: u) n = 0} R
3
.
where n = 0 is a xed normal vector and (: u) n is a dot prod-
uct. Prove that a plane through u is a coset of a two-dimensional
subspace of R
3
.
127
Solution. If the origin (0. 0. 0) lies on a plane H, then u = 0 and
H = {: = (x. y. z) R
3
: (:. n) = 0}.
where n = (. . ) is a (nonzero) normal vector; that is, H is the
set of all vectors orthogonal to n. In the words of Example 4.5,
H = u

, and so H is a subspace of R
3
.
Here are three vectors in H:
p = (. . 0); q = (0. . ); r = (. 0. ).
Since n = (. . ) = (0. 0. 0), at most one of these vectors
is 0. If p = 0 and q = 0, for example, it is easy to see that
p. q is linearly independent. Thus, p. q H, so that 2 =
dim( p. q) dim(H). Since H = R
3
, we have dim(H) - 3,
and so dim(H) = 2. By Proposition 4.22, p. q = H. (The
converse of this exercise is true, but we leave its proof for a linear
algebra course.)
4.11 (i) Prove that dim(Mat
mn
(k)) = mn.
Solution. A basis consists of the matrices E
i j
which have entry 1
in position i j and all other entries 0.
(ii) Determine dim(S), where S is the subspace of Mat
n
(k) consisting
of all the symmetric matrices.
Solution. n +
1
2
(n
2
n).
4.12 (i) Prove that the subset K of Mat
n
(k), consisting of all the skew
symmetric matrices, is a subspace of Mat
n
(k).
Solution. Now 0 is skew-symmetric, for 0
T
= 0 = 0. If A and
B are skew-symmetric, then
(A + B)
T
= A
T
+ B
T
= A B = (A + B).
and if is a scalar, then
(A)
T
= (A
T
) = A.
Therefore, K is a subspace of Mat
n
(k).
(ii) Determine dim(K).
Solution. If A is skew symmetric, then all its diagonal entries
are 0. The answer is
1
2
(n
2
n).
4.13 If p is a prime with p 1 mod 4, prove that there is a nonzero vector
: F
2
p
with (:. :) = 0, where (:. :) is the usual inner product of : with
itself [see Example 4.4(i)].
Solution. Use the Two-squares theorem.
128
4.14 Let k be a eld, and let k
n
have the usual inner product. Prove that if
: = a
1
e
1
+ +a
n
e
n
, then a
i
= (:. e
i
) for all i .
Solution. Prove that (e
i
. e
j
) =
i j
for all i. j , where
i j
is the Kronecker
delta.
4.15 If f (x) = c
0
+c
1
x + +c
m
x
m
k[x] and if A Mat
n
(k), dene
f (A) = c
0
I +c
1
A + +c
m
A
m
Mat
n
(k).
Prove that there is some nonzero f (x) k[x] with f (A) = 0.
Solution. Given A, then I. A. A
2
. . . . . A
m
2
is a linearly dependent list
because dim(Mat
m
(k)) = m
2
.
4.16 If U is a subspace of a vector space V over a eld k, then U is a subgroup
of V (viewed as an additive abelian group). Dene a scalar multiplication
on the cosets in the quotient group V,U by
(: +U) = : +U.
where k and : V. Prove that this is a well-dened function that
makes V,U into a vector space over k
Solution. Let : +U V,U and let be a scalar. If : +U = :

+U, then
: :

U. Since U is a subspace, (: :

) = : :

U, and so
: +U = :

+U. Thus, scalar multiplication on V,U is well-dened.

4.17 If V is a nite dimensional vector space and U is a subspace, prove that
dim(U) +dim(V,U) = dim(V).
Conclude that dim(V,U) = dim(V) dim(U).
Solution. Choose a basis u
1
. . . . . u
m
of U, so that dim(U) = m, and
extend it to a basis of V by adjoining vectors :
1
. . . . . :
r
. Show that :
1
+
U. . . . . :
r
+ U is a basis of V,U, and conclude that dim(V,U) = r =
dim(V) m.
4.18 Let Ax = b be a linear system of equations, and let s be a solution. If U
is the solution space of the homogeneous linear system Ax = 0, prove that
every solution of Ax = b has a unique expression of the form s + u for
u U. Conclude that the solution set of Ax = b is the coset s +U.
Solution. Absent.
4.19 If U and W are subspaces of a vector space V, dene
U + W = {u +n : u U and n W}.
(i) Prove that U + W is a subspace of V.
Solution. Absent.
129
(ii) If U and U

are subspaces of a nite dimensional vector space V,

prove that
dim(U) +dim(U

) = dim(U U

) +dim(U +U

).
Solution. Take a basis of U U

and extend it to bases of U and

of U

.
(iii) If V is nite dimensional, prove that every subspace U of V is a
direct summand.
Solution. Extend a basis B of U to a basis B C of V, and show
that V = U C.
4.20 (i) Prove that U W is a vector space.
Solution. Absent.
(ii) If U and W are nite dimensional vector spaces over a eld k,
prove that
dim(U W) = dim(U) +dim(W).
Solution. If B is a basis of U and C is a basis of W, then B C
is a basis of U V.
4.21 Assume that V is an n-dimensional vector space over a eld k and that V
has a nondegenerate inner product. If W is an r-dimensional subspace of
V, prove that V = WW

(see Example 4.5). Conclude that dim(W

) =
n r.
4.22 Here is a theoremof Pappus holding in k
2
, where k is a eld. Let and m be
distinct lines, let A
1
. A
2
. A
3
be distinct points on , and let B
1
. B
2
. B
3
be
distinct points on m. Dene C
1
to be A
2
B
3
A
3
B
2
, C
2
to be A
1
B
3
A
3
B
1
,
and C
3
to be A
1
B
2
A
2
B
1
. Then C
1
. C
2
. C
3
are collinear.
State the dual of the theorem of Pappus.
Solution. Let P and Q be distinct points, let
1
.
2
.
3
be distinct lines
passing through P, and let m
1
. m
2
. m
3
be distinct liness passing through
Q. Dene n
1
to be the line determined by
2
m
3
and
3
m
2
, n
2
to be the
line determined by
1
m
3
and
3
m
1
, and n
3
to be the line determined
by
1
m
2
and
2
m
1
. Then n
1
. n
2
. n
3
intersect in a point.
4.23 True or false with reasons.
(i) There is a solution to an n n inhomogeneous system Ax = b if
A is a triangular matrix.
Solution. False.
(ii) Gaussian equivalent matrices have the same row space.
Solution. True.
130
(iii) Gaussian equivalent matrices have the same column space.
Solution. False.
(iv) The matrix A =
_
1 0 0
0 1 1
0 0 1
_
is nonsingular.
Solution. True.
(v) Every nonsingular matrix over a eld is a product of elementary
matrices.
Solution. True.
(vi) If A is an m n matrix, then Ron(A
T
) = Col(A).
Solution. True.
4.24 (i) Prove that a list :
1
. . . . . :
m
in a vector space V is linearly inde-
pendent if and only if it spans an m-dimensional subspace of V.
Solution.
(ii) Determine whether the list
:
1
= (1. 1. 1. 2). :
2
= (2. 2. 3. 1). :
3
= (1. 1. 0. 5)
in k
4
is linearly independent.
Solution. Let A be the matrix whose rows are the given vectors,
and see whether rank(A) = m.
4.25 Do the vectors :
1
= (1. 4. 3), :
2
= (1. 2. 0), :
3
= (2. 2. 3) span k
3
?
Solution. Yes, because rank(A) = 3.
4.26 (i) Prove that every n n row reduced echelon matrix is triangular.
Solution. Absent.
(ii) Use Theorem 4.39 to prove that every n n matrix A is Gaussian
equivalent to a triangular matrix.
Solution. Absent.
4.27 Prove that Ax = is consistent if and only if lies in the column space
of A.
Solution. If k
m
, then A is a linear combination of the columns of A.
4.28 If A is an n n nonsingular matrix, prove that any system Ax = b has a
unique solution, namely, x = A
1
b.
Solution. Absent.
4.29 Let
1
. . . . .
n
be the columns of an m n matrix A over a eld k, and let
k
m
.
(i) Prove that
1
. . . . .
n
if and only if the inhomogeneous
system Ax = has a solution.
Solution. Absent.
131
(ii) Prove that lies in the column space of A if and only if rank([A|])
= rank(A).
Solution. Let A be Gaussian equivalent to an echelon matrix U,
so that there is a nonsingular matrix P with PA = U. Then lies
in the row space Ron(A) if and only if P Ron(U).
(iii) Does = (0. 3. 5) lie in the subspace spanned by
1
= (0. 2. 3),

2
= (0. 4. 6),
3
= (1. 1. 1)?
Solution. No.
4.30 (i) Prove that an n n matrix A over a eld k is nonsingular if and
only if it is Gaussian equivalent to the identity I .
Solution. Absent.
(ii) Find the inverse of
A =
_
_
2 3 1
1 1 0
1 0 1
_
_
.
Solution. If E
p
E
1
A = I , then A
1
= E
1
1
E
1
p
. Con-
clude that the elementary row operations which change A into I
also change I into A
1
. The answer is
A
1
=
1
4
_
_
1 3 1
1 1 1
1 3 5
_
_
.
4.31 (i) Let Ax = b be an m n linear system over a eld k. Prove that
there exists a solution x = (x
1
. . . . . x
n
) with x
j
1
= 0 = x
j
2
=
= x
j
s
, where s n, if and only if there is a solution to the
m (n s) system A

= b, where A

is obtained from A be
deleting columns j
1
. j
2
. . . . . j
s
from A.
Solution. Absent.
(ii) Prove that if the matrix A

in part (i) has rank m, then there exists

a solution to Ax = b with x
j
1
= 0 = x
j
2
= = x
j
s
.
Solution. Absent.
4.32 True or false with reasons.
(i) Every linear transformation T : V V, where V is a nite di-
mensional vector space over R, is represented by innitely many
matrices.
Solution. False.
132
(ii) Every matrix over R is similar to innitely many different matri-
ces.
Solution. False.
(iii) If S and T are linear transformations on the plane R
2
that agree on
two nonzero points, then S = T.
Solution. True.
(iv) If A and B are n n nonsingular matrices, then A + B is nonsin-
gular.
Solution. False.
(v) If A and B are nn nonsingular matrices, then AB is nonsingular.
Solution. True.
(vi) If k is a eld, then
{A Mat
n
(k) : AB = BA for all B Mat
n
(k)}
is a 1-dimensional subspace of Mat
n
(k).
Solution. True.
(vii) The vector space of all 3 3 symmetric matrices over R is iso-
morphic to the vector space consisting of 0 and all f (x) R[x]
with deg( f ) 5.
Solution. True.
(viii) If X and Y are bases of a nite dimensional vector space over a
eld k, then
Y
[1
V
]
X
is the identity matrix.
Solution. False.
(ix) Transposition Mat
mn
(C) Mat
nm
(C), given by A A
T
, is
a nonsingular linear transformation.
Solution. True.
(x) If V is the vector space of all continuous f : [0. 1] R, then
integration f
_
1
0
f (x) dx is a linear transformation V R.
Solution. True.
4.33 Let k be a eld, let V = k[x], the polynomial ring viewed as a vector space
over k, and let V
n
= 1. x. x
2
. . . . . x
n
. By Exercise 4.8, we know that
X
n
= 1. x. x
2
. . . . . x
n
is a basis of V
n
.
(i) Prove that differentiation T : V
3
V
3
, dened by T( f (x)) =
f

(x), is a linear transformation, and nd the matrix A =
X
3
[T]
X
3
of differentiation.
Solution. Absent.
133
(ii) Prove that integration S : V
3
V
4
, dened by S( f ) =
_
x
0
f (t ) dt ,
is a linear transformation, and nd the matrix A =
X
4
[S]
X
3
of in-
tegration.
Solution. Absent.
4.34 If S
n
and P = P

is the corresponding permutation matrix, prove that

P
1
= P
T
.
Solution. Absent.
4.35 Let V and W be vector spaces over a eld k, and let S, T : V W be
linear transformations.
(i) If V and W are nite dimensional, prove that
dim(Hom
k
(V. W)) = dim(V) dim(W).
Solution. Absent.
(ii) If X = :
1
. . . . . :
n
is a basis of V, dene
1
. . . . .
n
V

by

i
(:
j
) =
_
0 if j = i
1 if j = i .
Prove that
1
. . . . .
n
is a basis of V

.
Solution. Absent.
(iii) If dim(V) = n, prove that dim(V

) = n, and hence that V

= V.
Solution. Absent.
4.36 (i) If S : V W is a linear transformation and f W

, then the
composite V
S
W
f
k lies in V

. Prove that S

: W

,
dened by S

: f f S, is a linear transformation.
Solution. Absent.
(ii) If X = :
1
. . . . . :
n
and Y = n
1
. . . . . n
m
are bases of V and W,
respectively, denote the dual bases by X

and Y

. If S : V W is
a linear transformation, prove that the matrix of S

is a transpose:
X
[S

]
Y
=
_
Y
[S]
X
_
T
.
Solution. Absent.
4.37 (i) If A =
_
a b
c d
_
, dene det(A) = ad bc. Given a system of linear
equations Ax = 0 with coefcients in a eld,
ax +by = p
cx +dy = q.
134
prove that there exists a unique solution if and only if det(A) = 0.
Solution. Absent.
(ii) If V is a vector space with basis X = :
1
. :
2
, dene T : V V
by T(:
1
) = a:
1
+b:
2
and T(:
2
) = c:
1
+d:
2
. Prove that T is a
nonsingular linear transformation if and only if det(
X
[T]
X
) = 0.
Solution. Absent.
4.38 Let U be a subspace of a vector space V.
(i) Prove that the natural map : V V,U, given by : : +U,
is a linear transformation with kernel U.
Solution. Absent.
(ii) State and prove the rst isomorphism theorem for vector spaces.
Solution. Here is the statement. If f : V W is a linear trans-
formation with ker f = U, then U is a subspace of V and there is
an isomorphism : V,U im f , namely, (: +U) = f (:).
4.39 Let k be a eld and let k

be its multiplicative group of nonzero ele-

ments. Prove that det : GL(2. k) k

is a surjective group homomor-

phism whose kernel is SL(2. k). Conclude that SL(2. k) GL(2. k) and
GL(2. k), SL(2. k)

= k

.
Solution. Absent.
4.40 Let V be a nite dimensional vector space over a eld k, and let B denote
the family of all the bases of V. Prove that B is a transitive GL(V)-set.
Solution. Absent.
4.41 (i) Let U = {(a. a) : a R}. Find all the complements of U in R
2
.
Solution. If : = (b. c) is any nonzero vector with b = c, then
W
:
= {a: : a R} is a complement of U. Moreover, every
complement is of this form.
(ii) If U is a subspace of a nite dimensional vector space V, prove
that any two complements of U are isomorphic.
Solution. Absent.
4.42 If A is an m n matrix and B is an p m matrix, prove that
rank(BA) rank(A).
Solution. Absent.
4.43 Let R
n
be equipped with the usual inner product: if : = (a
1
. . . . . a
n
) and
u = (b
1
. . . . . b
n
), then (:. u) = a
1
b
1
+ +a
n
b
n
.
(i) Prove that every orthogonal transformation is nonsingular.
Solution. Absent.
135
(ii) Prove that a linear transformation U : R
n
R
n
is orthogonal if
and only if U(:
1
). . . . . U(:
n
) is an orthonormal basis whenever
:
1
. . . . . :
n
is an orthonormal basis.
Solution. Absent.
(iii) If n R
n
and :
1
. . . . . :
n
is an orthonormal basis, then n =

n
i =1
c
i
:
i
. Prove that c
i
= (n. :
i
).
Solution. Absent.
4.44 Let U : R
n
R
n
be an orthogonal transformation, and let X = :
1
. . . . . :
n
be an orthonormal basis. If O =
X
[U]
X
, prove that O
1
= O
T
.
Solution. Absent.
4.45 Let A be an n n real symmetric matrix.
(i) Give an example of a nonsingular matrix P for which PAP
1
is
not symmetric.
Solution. Absent.
(ii) Prove that OAO
1
is symmetric for every n n real orthogonal
matrix O.
Solution. Absent.
4.46 True or false with reasons.
(i) If a matrix A is similar to a symmetric matrix, then A is symmetric.
Solution. False.
(ii)
_
2 1
0 1
_
is invertible over Q.
Solution. True.
(iii)
_
2 1
0 1
_
is invertible over Z.
Solution. False.
(iv) If A is a 2 2 matrix over R all of whose entries are positive, then
det(A) is positive.
Solution. False.
(v) If A is a 2 2 matrix over R all of whose entries are positive, then
det(A) 0.
Solution. False.
(vi) If A and B are n n matrices, then tr(A + B) = tr(A) +tr(B).
Solution. True.
(vii) If two n n matrices over a eld k have the same characteristic
polynomial, then they are similar.
Solution. False.
136
(viii) If A =
_
1 2
3 4
_
, then A
2
5A 2I = 0.
Solution. True.
(ix) Every n n matrix over R has a real eigenvalue.
Solution. False.
(x)
_
2 1 7
0 1 8
0 0 0
_
is diagonalizable.
Solution. True.
4.47 Let R be a commutative ring, let D: Mat
n
(R) R be a determinant
function, and let A be an n n matrix with rows
1
. . . . .
n
. Dene
d
i
: R
n
R by d
i
() = D(
i
. . . . .
i 1
. .
i +1
. . . . .
n
).
(i) If i = j and r R, prove that
d
i
(r
j
) = 0.
Solution. Absent.
(ii) If i = j and r R, prove that d
i
(
i
+r
j
) = D(A).
Solution. Absent.
(iii) If r
j
R, prove that
d
i
(
i
+

j =i
r
j

j
) = D(A).
Solution. Absent.
4.48 If O is an orthogonal matrix, prove that det(O) = 1.
Solution. Absent.
4.49 If A

is obtained from an n n matrix by interchanging two of its rows,

prove that det(A

) = det(A).
Solution. Absent.
4.50 If A is an n n matrix over a commutative ring R and if r R, prove that
det(r A) = r
n
det(A). In particular, det(A) = (1)
n
det(A).
Solution. Absent.
4.51 If A = [a
i j
] is an n n triangular matrix, prove that
det(A) = a
11
a
22
a
nn
.
Solution. Induction on n 1.
4.52 If u
1
. . . . . u
n
is a list in a eld k, then the corresponding Vandermonde
matrix is the n n matrix
V = Van(u
1
. . . . . u
n
) =
_
_
_
_
_
_
_
1 u
1
u
2
1
u
3
1
u
n1
1
1 u
2
u
2
2
u
3
2
u
n1
2
.
.
.
.
.
.
.
.
.
.
.
.
1 u
n
u
2
n
u
3
n
u
n1
n
_

_
.
137
(i) Prove that
det(V) =

i -j
(u
j
u
i
).
Conclude that V is nonsingular if all the u
i
are distinct.
Solution. The proof is by induction on n 1, but we merely
describe the inductive step from n = 3 to n = 4. Let
V =
_
_
_
_
_
_
1 a a
2
a
3
1 b b
2
b
3
1 c c
2
c
3
1 d d
2
d
3
_

_
.
Subtract the top row from each of the other rows (which does not
change the determinant):
det(V) = det
_
_
_
_
_
_
_
_
_
_
_
_
1 a a
2
a
3
0 b a b
2
a
2
b
3
a
3
0 c a c
2
a
2
c
3
a
3
0 d a d
2
a
2
d
3
a
3
_

_
_
_
_
_
_
_
.
By Laplace expansion across the top row, we have
det(V) = det
_
_
_
_
_
_
b a b
2
a
2
b
3
a
3
c a c
2
a
2
c
3
a
3
d a d
2
a
2
d
3
a
3
_

_
_
_
_.
Use the identities x
2
y
2
= (x y)(x + y) and x
3
y
3
=
(x y)(x
2
+xy +y
2
), and factor out b a, c a, and d a from
rows 1, 2, and 3, respectively, to obtain
det(V) = (ba)(ca)(da) det
_
_
_
_
_
_
1 b +a b
2
+ba +a
2
1 c +a c
2
+ca +a
2
1 d +a d
2
+da +a
2
_

_
_
_
_.
Subtract aCOL(2) from COL(3), and then subtract aCOL(1) from
COL(2) to obtain
det(V) = (b a)(c a)(d a) det
_
_
_
_
_
_
1 b b
2
1 c c
2
1 d d
2
_

_
_
_
_.
The last matrix is a 3 3 Vandermonde matrix.
138
(ii) If is a primitive nth root of unity (
n
= 1 and
i
= 1 for i - n),
prove that Van(1. .
2
. . . . .
n1
) is nonsingular and that
Van(1. .
2
. . . . .
n1
)
1
=
1
n
Van(1.
1
.
2
. . . . .
n+1
).
Solution. Absent.
(iii) Let f (x) = a
0
+ a
1
x + a
2
x
2
+ + a
n
x
n
k[x], and let y
i
=
f (u
i
). Prove that the coefcient vector a = (a
0
. . . . . a
n
) is a
solution of the linear system
Vx = y. (4)
where y = (y
0
. . . . . y
n
). Conclude that if all the u
i
are distinct,
then f (x) is determined by Eq. (4).
Solution. Absent.
4.53 Let T[x
1
. . . . . x
n
] be an n n tridiagonal matrix.
(i) If D
n
= det(T[x
1
. . . . . x
n
]), prove that D
1
= x
1
, D
2
= x
1
x
2
+1,
and, for all n > 2,
D
n
= x
n
D
n1
+ D
n2
.
Solution. Absent.
(ii) Prove that if all x
i
= 1, then D
n
= F
n+1
, the nth Fibonacci num-
ber.
Solution. Absent.
4.54 Let T : C
n
C
n
be a linear transformation, and let A = [a
i j
] be its matrix
relative to the standard basis e
1
. . . . . e
n
.
(i) If ( . ) is the hermitian form on C
n
, prove that there exists a linear
transformation T
#
: C
n
C
n
with
(Tu. :) = (u. T
#
:)
for all u. : C
n
.
Solution. Absent.
(ii) Prove that the matrix of T
#
relative to the standard basis is A
#
=
[a
j i
].
Solution. Absent.
(iii) A real nn matrix A denes a linear transformation T : C
n
C
n
by matrix multiplication: T(:) = A:. Prove that if A is symmet-
ric, then T
#
= T.
Solution. Absent.
139
(iv) Prove that every hermitian matrix A over C is diagonalizable.
Solution. Absent.
4.55 If A is an m n matrix over a eld k, prove that rank(A) d if and
only if A has a nonsingular d d submatrix. Conclude that rank(A) is the
maximum such d.
Solution. Absent.
4.56 (i) If A and B are n n matrices with entries in a commutative ring
R, prove that tr(AB) = tr(BA).
Solution. Absent.
(ii) Using part (i) of this exercise, give another proof of Corollary 4.96:
if A and B are similar matrices with entries in a eld k, then
tr(A) = tr(B).
Solution. Absent.
4.57 If A is an nn matrix over a eld k, where n 2, prove that det(adj(A)) =
det(A)
n1
.
Solution. Absent.
4.58 If C = C(g) is the companion matrix of g(x) k[x], prove that the
characteristic polynomial h
C
(x) = det(x I C) = g(x).
Solution. Absent.
4.59 Let R be a commutative ring. If A
1
. . . . . A
t
are square matrices over R,
prove that
det(A
1
A
t
) = det(A
1
) det(A
t
).
Solution. Absent.
4.60 If A
1
. . . . . A
t
and B
1
. . . . . B
t
are square matrices with A
i
similar to B
i
for
all i , prove that A
1
A
t
is similar to B
1
B
t
.
Solution. Absent.
4.61 Prove that an n n matrix A with entries in a eld k is singular if and only
if 0 is an eigenvalue of A.
Solution. Absent.
4.62 Let A be an n n matrix over a eld k. If c is an eigenvalue of A, prove,
for all m 1, that c
m
is an eigenvalue of A
m
.
Solution. Absent.
4.63 Find all possible eigenvalues of n n matrices A over R for which A and
A
2
are similar.
Solution. Absent.
4.64 Prove that all the eigenvalues of a nilpotent matrix are 0. Use the Cayley-
Hamilton theorem to prove the converse: if all the eigenvalues of a matrix
A are 0, then A is nilpotent.
140
Solution. If A: = c:, then A
m
: = c
m
: for all m 1. Hence, if A
m
= 0,
then c
m
= 0 (because eigenvectors are nonzero). Thus, c = 0.
Conversely, if all the eigenvalues of an n n matrix A are 0, then the
characteristic polynomial is h
A
(x) = (x 0) (x 0) = x
n
. By the
Cayley-Hamilton theorem, A
n
= 0, and so A is nilpotent.
4.65 If N is a nilpotent matrix, prove that I + N is nonsingular.
Solution. Dene N
1
= I N +N
2
N
3
+ +N
m1
, where N
m
= 0.
4.66 Let Abe an alphabet with |A| = q 2, let T : A
n
A
n
be a transmission
function, and let the probability of error in each letter of a transmitted word
be p, where 0 - p - 1.
(i) Prove that the probability P of the occurrence of exactly erro-
neous letters in a transmitted word of length n is
P =
_
p
q 1
_

(1 p)
n
.
Solution. Absent.
(ii) Prove that P =
_
n

_
p

(1 p)
n
, and conclude that the probabil-
ity P is independent of q.
Solution. Absent.
4.67 Prove that d 3, where d is the minimum distance of the two-dimensional
parity code in Example 4.105(iii).
Solution. Absent.
4.68 Let A be an alphabet with |A| = q, and let C A
n
be an (n. M. d)-code.
(i) Prove that the projection : C A
nd+1
, dened by
(c
1
. . . . . c
n
) = (c
d
. . . . . c
n
).
is an injection.
Solution. Let c = (c
1
. . . . . c
n
) and c

= (c

1
. . . . . c

n
) lie in C. If
(c) = (c

), then (c
d
. . . . . c
n
) = (c

d
. . . . . c

n
); that is, c and c

agree in at least n d +1 positions, and so (c. c

) d 1 - d.
Since d is the minimum distance, c = c

and is an injection.
(ii) (Singleton bound.) Prove that M q
nd+1
.
Solution. M = |C|, and part (i) gives |C| q
nd+1
.
4.69 (i) If A is an alphabet with |A| = q and is the Hamming distance,
prove that

_
n A
n
: (u. n) = i
_

=
_
n
i
_
(q 1)
i
.
141
Solution. Absent.
(ii) Prove that
|B
r
(u)| =
r

i =0
_
n
i
_
(q 1)
i
.
Solution. Absent.
(iii) (Gilbert-Varshamov bound.) If C A
n
is an (n. M. d)-code,
where |A| = q, prove that
q
n

d1
i =0
_
n
i
_
(q 1)
i
M.
Solution. Absent.
4.70 (Hamming bound)If C A
n
is an (n. M. d)-code, where |A| = q and
d = 2t +1, prove that
M
q
n

t
i =0
_
n
i
_
(q 1)
i
.
Solution. Absent.
4.71 Prove that the Hamming [2

1. 2

1 ] codes in Example 4.113 are

perfect codes.
Solution. Absent.
4.72 Suppose that a code C detects up to s errors and that it corrects up to t
errors. Prove that t s.
Solution. Absent.
4.73 If C F
n
is a linear code and n F
n
, dene r = min
cC
(n. c). Give
an example of a linear code C F
n
, which corrects up to t errors, and a
word n F
n
with n , C, such that there are distinct codewords c. c

C
with (n. c) = r = (x. c

). Conclude that correcting a transmitted word

by choosing the codeword nearest to it may not be well-dened.
Solution. If C = {[a. a] F
2
}, then C corrects up to 1 error. If n = [a. b],
where a = b, then n , C and (n. [a. a]) = 1 = (n. [b. b]).
4.74 Let C be an [n. m] linear code over a nite eld F, and let G be a generating
matrix of C. Prove that an m n matrix A is also a generating matrix of C
if and only if A = GH for some matrix H GL(n. F).
Solution. Absent.
4.75 Prove that the BCH code of length m +1 over F
2
having generating poly-
nomial x 1 is the [m +1. m] parity check code.
Solution. Absent.
142
4.76 (i) Write x
15
1 as a product of irreducible polynomials in F
2
[x].
Solution. Absent.
(ii) Find an irreducible quartic polynomial g(x) F
2
[x], and use it to
dene a primitive 15th root of unity F
16
.
Solution. Absent.
(iii) Find a BCHcode C over F
2
of length 15 having minimumdistance
d(C) 3.
Solution. Absent.
4.77 Let C be the 2-error correcting Reed-Solomon [7. 3] code over F
8
in Ex-
ample 4.131. Decode the word y = [
3
. . 1.
3
+ . 0.
3
.
3
] assuming
that it has an error vector of weight 2.
Solution. Use Table 4.1 in Example 4.127. The answer is
c = [
3
. . 1 +
3
.
3
+. 0.
3
. 1].
Exercises for Chapter 5
5.1 (i) Find the roots of f (x) = x
3
3x +1.
Solution. Here, q = 3, r = 1, and D = 27r
2
+4q
3
= 81 - 0.
Using Vi` etes formula, we have t = 2 and 3 = 120

, so that the
roots are
2 cos 40

. 2 cos 160

. 2 cos 280

.
(ii) Find the roots of f (x) = x
3
9x +28.
Solution. 4. 2 i

3.
(iii) Find the roots of f (x) = x
3
24x
2
24x 25.
Solution. 17.
1
2
i

3
2
.
5.2 (i) Find the roots of f (x) = x
3
15x 4 using the cubic formula.
Solution. g =
3
_
2 +

121 and h =
3
_
2

121.
(ii) Find the roots of f (x) using the trigonometric formula.
Solution. 4. 2

3.
5.3 Find the roots of f (x) = x
3
6x +4.
Solution. 2. 1

3.
5.4 Find the roots of x
4
15x
2
20x 6.
Solution. 3. 1. 2

6.
143
5.5 There is a circular castle whose diameter is unknown; it is pro-
vided with four gates, and two lengths out of the north gate there
is a large tree, which is visible from a point six lengths east of
the south gate. What is the length of the diameter?
S E
N
C
O
T
2
r
r
r
a
6
Figure 5.7 The Castle Problem
(i) Prove that the radius r of the castle is a root of the cubic X
3
+
X
2
36.
Solution. We compute the area of LT SE in two ways. On the
one hand,
area(LT SE) =
1
2
(2 +2r)6 = 6 +6r.
On the other hand, this area is the sum of the areas of the three
smaller triangles: LOSE; LOCE; LOCT. Now the right tri-
angles LOSE and LOCE are congruent (they have same hy-
potenuse and a leg of equal length, namely, r), and so they have
the same area:
1
2
6r = 3r; also, area(LOSE) =
1
2
ra, where
a = |CT|. We conclude that 6 + 6r = 3r + 3r +
1
2
ra; that
is,
12 = ra.
The Pythagorean theorem (applied to LOCT) gives
r
2
+a
2
= (r +2)
2
= r
2
+4r +4.
and so a
2
= 4r +4. Since 12 = ra, we have
144
r
2
= 4r +4.
and this simplies to r
3
+r
2
36 = 0.
144
(ii) Show that one root of f (X) = X
3
+ X
2
36 is an integer and
nd the other two roots. Compare your method with Cardanos
formula and with Vi` etes trigonometric solution.
Solution. Corollary 3.91 says that any integer root is a divisor of
36, and one checks that 3 is a root; therefore,
X
3
+ X
2
36 = (X 3)(X
2
+4X +12).
The quadratic formula gives the other two roots: 4

8, both
of which are negative.
The other two methods are longer. Both require the substitution
X = x
1
3
, yielding the reduced cubic x
3

1
3
x
970
27
. We do not
give the calculations.
5.6 Show that if u is a root of a polynomial f (x) R[x], then the complex
conjugate u is also a root of f (x).
Solution. Let f (x) =

r
i
x
i
. If 0 =

r
i
u
i
, then since complex conju-
gation is an isomorphism xing each real number,
0 = 0 =

r
i
u
i
=

r
i
u
i
= f (u).
5.7 Assume that 0 3 - 360

.
(i) If cos 3 is positive, show that there is an acute angle with 3 =
3 or 3 = 3( +90

), and that the sets of numbers

cos . cos( +120

). cos( +240

)
and
cos( +90

). cos( +210

). cos( +330

)
coincide.
(ii) If cos 3 is negative, show that there is an acute angle with 3 =
3( +30

) or 3 = 3( +60

), and that the sets of numbers

cos( +30

). cos( +150

). cos( +270

)
and
cos( +60

). cos( +180

). cos( +270

)
coincide.
5.8 Show that if cos 3 = r, then the roots of 4x
3
3x r are
cos . cos( +120

). and cos( +240

).
145
Solution. We know that cos is a root of f (x) = 4x
3
3x r, where
r = cos 3. In particular, if = + 120

, then cos is a root of 4x

3

3x cos 3( +120

). But the addition formula for cosine gives

cos 3( +120

) = cos 3 = r.
and so cos( +120

) is also a root of f (x). Similarly, cos 3( +240

) =
cos3, and cos( +240

) is also a root of f (x).

5.9 (i) Prove that cosh(3) = 4 cosh
3
() 3 cosh().
Solution. By denition, cosh =
1
2
(e

+ e

). Expand and then

simplify
4[
1
2
(e

+e

)]
3
3[
1
2
(e

+e

)]
to obtain
1
2
(e
3
+e
3
).
(ii) Prove that sinh(3) = 4 sinh
3
() +3 sinh().
Solution. Absent.
5.10 Find the roots of x
3
9x +28.
Solution. The cubic formula works smoothly: q = 9, r = 28, D = 676,
= 1, = 3, and the roots are 4 and 2

3.
5.11 Find the roots of x
3
24x
2
24x 25.
Solution. After substituting X = x + 8, one obtains x
3
216x 1241.
For the new cubic, q = 216, r = 1241, D = 47089 = (217)
2
, = 9,
= (216),3 8, and so a root is 17; it follows that a root of the original
cubic is 17 + 8 = 25. The division algorithm gives
X
3
24X
2
24X 25 = (X 25)(X
2
+ X +1).
and so the other two roots are
1
2
(1

3).
5.12 (i) Find the roots of x
3
15x 4 using the cubic formula.
Solution. We have q = 15, r = 4, D = 484, and
3
=
2 +

121. It is not clear how to use these formulas to give
numerical estimates for the roots.
(ii) Find the roots using the trigonometric formula.
Solution. We have t =

20 4.47 and cos 3 .179. Hence
3 80

and 27

. Hence
t cos

20 cos 27

4.47 .891 3.98.

Actually, one root is 4, as one sees using Corollary 3.91. It is
simplest to use the division algorithm. Since
x
3
15x
2
4 = (x 4)(x
2
+4x +1).
the other two roots are 2

3.
146
5.13 Find the roots of x
3
6x +4.
Solution. We have q = 6, r = 4, D = 16, and
3
= 2 +2i . Use De
Moivres theorem to nd cube roots of 2 +2i :
(1 +i )
3
= 2 +2i.
Hence, = 1 +i , = 1 i , and a root of the polynomial is + = 2.
The other two roots can be found by two methods. We can use the cubic
formula to obtain:
(1 +i ) +
2
(1 i ) and
2
(1 +i ) +(1 i ).
or one can use the division algorithm and the quadratic formula:
x
3
6x +4 = (x 2)(x
2
+2x 2).
so that the other roots are 1

3. It is not clear from the cubic for-
mula that the roots are all real; however, 27r
2
+ 4q
3
= 432 - 0, and
Proposition 5.9 applies.)
5.14 Find the roots of x
4
15x
2
20x 6.
Solution. This is a realistic problem; the implementation of the formula is
rather long (but see the remark at the end of this solution).
Set Q = 15, R = 20, and S = 6. Then
j
6
+2Qj
4
+(Q
2
4S) j
2
R
2
= j
6
30 j
4
+249 j
2
400.
We now make this cubic (in j
2
) reduced with the substitution j
2
= y +10,
obtaining
F(y) = y
3
51y +90.
Apply the cubic formula: q = 51, r = 90, D = 11552. Thus,

D
107.5i ,
g
3

1
2
(90 +107.5i ) 45 +53.8i.
and h
3
45 53.8i . Since g
3
and h
3
are complex conjugates, both g
and h can be chosen to be complex conjugates: if g
3
= z, then g
3
= z; it
follows that g+h = g+g is real. We now use De Moivres theorem to take
the cube root of these complex numbers: g 3+2.8i and h 32.8i , so
that y = g +h = 6 is a root of F(y) (alternatively, one could have checked
for integer roots of F(y) using Corollary 3.91). Therefore,
j
2
= 16
(for j
2
= y + 10) and j = 4. Eqs. (10) in the proof of Theorem 5.10 are
thus
2m = 16 15
20
4
= 4
2 = 16 15 +
20
4
= 6.
147
Thus, m = 2, = 3, and we have the factorization
x
4
15x
2
20x 6 = (x
2
+4x +3)(x
2
4x 2).
The quadratic formula applied to each of the two factors gives the desired
roots:
3. 1. 2 +

6. 2

6.
In particular, all the roots are real.
Remark. Given the discussion of roots in Chapter 3, especially Corol-
lary 3.91, it is natural to look for rational, hence integral roots, and to nd
that 1 and 3 are roots (the candidates are 1, 2, 3, and 6). There-
fore, long division gives
x
4
15x
2
20x 6 = (x +1)(x +3)(x
2
4x 2).
The other two roots are thus 2 +

6 and 2

6.
We now understands the unpopularity of the quartic formula.
5.15 True or false with reasons.
(i) Every algebraically closed eld contains n distinct nth roots of
unity, where n 1.
Solution. False.
(ii) There are no 5th roots of unity in a eld of characteristic 5.
Solution. False.
(iii) R is a splitting eld of x
2
5 over Q.
Solution. False.
(iv) Q(

5) is a normal extension of Q.
Solution. False.
(v) No polynomial of degree 5 in Q[x] is solvable by radicals.
Solution. False.
(vi) F
2
(x) = Frac(F
2
[x]) is an innite eld of characteristic 2.
Solution. True.
(vii) A polynomial f (x) Q[x] can have two splitting elds inside of
C.
Solution. False.
(viii) The alternating group A
4
is a solvable group.
Solution. True.
148
(ix) The alternating group A
5
is a solvable group.
Solution. False.
5.16 Let : A H be a group homomorphism. If B A and B ker ,
prove that the induced map

: A,B H, given by aB (a), is a

well-dened homomorphism with im

= im.
Solution. If aB = cB, then c
1
a B ker , and so (c
1
a) = 1.
Therefore, (c) = (a), and

is well-dened.
5.17 If z C is a constructible number, prove that Q(i. z),Q is a radical exten-
sion.
Solution. Absent.
5.18 Let k be a eld and let f (x) k[x]. Prove that if E and E

are splitting
elds of f (x) over k, then Gal(E,k)

= Gal(E

,k).
Solution. Absent.
5.19 Prove that F
3
[x],(x
3
x
2
1)

= F
3
[x],(x
3
x
2
+ x 1).
Solution. We saw, in Chapter 3, that both cubics are irreducible in F
3
[x],
and so the quotient rings are elds of order 3
3
= 27. But any two nite
elds of the same order are isomorphic.
5.20 Is F
4
a subeld of F
8
?
Solution. If F
4
were a subeld of F
8
, then F

4
would be a subgroup of F

8
.
But this contradicts Lagranges theorem, for 3 7.
5.21 Let k be a eld of characteristic p > 0, and dene the Frobenius map
F : k k by F : a a
p
.
(i) Prove that F : k k is an injection.
Solution. This follows from ker F = {a k : a
p
= 0} = {0}.
(ii) When k is nite, prove that F is an automorphism xing the prime
eld F
p
. Conclude that F Gal(k,F
p
).
Solution. It is clear that F(1) = 1 and F(ab) = F(a)F(b). To
prove F(a + b) = F(a) + F(b), that is, (a + b)
p
= a
p
+ b
p
,
use the binomial theorem and Proposition 1.39, which says that
p |
_
p
i
_
if 0 - i - p.
That F xes F
p
is just a restatement of Fermats theorem.
(iii) Prove that if k is nite, then every a k has a pth root; that is,
there is b k with b
p
= a.
Solution. We know that the Frobenius F : k k is injective;
since k is nite, it must also be surjective, by Exercise 2.13. But
this says that every element in k has a pth root.
5.22 Let q = p
n
for some prime p and some n 1.
(i) If is a generator of F

q
, prove that F
q
= F
p
().
149
Solution. If K,k is a eld extension and K, then k() is the
subeld of K generated by k and . In particular, F
p
() = F
q
because every nonzero element in F
q
is a power of .
(ii) Prove that the irreducible polynomial p(x) F
p
[x] of has de-
gree n.
Solution. If deg( p) = d, then Proposition 4.30 says that
[F
p
() : F
p
] = d; it follows that |F
p
()| = p
d
. But |F
p
()| =
|F
p
n | = p
n
, and so d = n.
(iii) Prove that if G = Gal(F
q
,F
p
), then |G| n.
Solution. Every nonzero element of F
q
has the form
i
for some
i , and (
i
) = [()]
i
. Now permutes the roots of p(x), the
irreducible polynomial of . Since deg( p) = n, there are at most
n choices for (), and so |G| n.
(iv) Prove that Gal(F
q
,F
p
) is cyclic of order n with generator the
Frobenius F.
Solution. If F
j
= 1 for some j - n, then a
p
j
= a for all a F
q
.
This says that the polynomial x
p
j
x has p
n
> p
j
roots, and this
is a contradiction.
5.23 Given f (x) = ax
2
+ bx + c Q[x], prove that the following statements
are equivalent.
(i) f (x) is irreducible.
(ii)

b
2
4ac is not rational.
(iii) Gal(Q(

b
2
4ac),Q) has order 2.
Solution.
(i) (ii). A quadratic is irreducible if and only if it has no rational roots.
By the quadratic formula,

b
2
4ac is not rational.
(ii) (iii). If

b
2
4ac is irrational, then the splitting eld of f (x)
is Q(

b
2
4ac), which is a proper extension eld of Q. There-
fore, | Gal(Q(

b
2
4ac)| 2. On the other hand, we know that
Gal(Q(

b
2
4ac) is isomorphic to a subgroup of S
2

= F
2
, by The-
orem 5.21, and so | Gal(Q(

b
2
4ac)| = 2.
(iii) (i). If f (x) factors in Q[x], then f (x) = (x a)(x b) for
a. b Q. Therefore, Q is the splitting eld of f (x) and so the Galois
group has order 1.
5.24 Let E,k be a splitting eld of a polynomial f (x) k[x]. If deg( f ) = n,
prove that [E : k] n!. Conclude that E,k is a nite extension.
Solution. Absent.
150
5.25 What is the degree of the splitting eld of x
30
1 over F
5
?
Solution. Observe that x
30
1 = (x
6
1)
5
in F
5
[x].
5.26 Prove that if f (x) Q[x] has a rational root a, then its Galois group is the
same as the Galois group of f (x),(x a).
Solution. If f (x) = (x a)g(x), then the splitting eld of f (x) over Q
is the same as the splitting eld of g(x); hence, the Galois groups of f (x)
and of g(x) are equal.
5.27 (i) Let H be a normal subgroup of a nite group G. If both H and
G,H are solvable groups, prove that G is a solvable group.
Solution. Absent.
(ii) If H and K are solvable groups, prove that H K is solvable.
Solution. Absent.
5.28 We are going to improve Theorem 5.34 by eliminating the hypothesis in-
volving roots of unity: if k is a eld and f (x) k[x] is solvable by radicals,
then its Galois group Gal(E,k) is a solvable group.
Since f (x) is solvable by radicals, there is a radical tower k = K
0

F with E F; moreover, we were able to assume that F,k a
splitting eld of some polynomial. Finally, if k contains a certain set O of
mth roots of unity, then Gal(E,k) is solvable.
(i) Dene E

,E to be a splitting eld of x
m
1, and dene k

=
k(O). Prove that E

is a splitting eld of f (x) over k

, and con-
clude that Gal(E

,k

) is solvable.
Solution. Absent.
(ii) Prove that Gal(E

,k

) Gal(E

,k) and
Gal(E

,k), Gal(E

,k

)

= Gal(k

,k).
Solution. Absent.
(iii) Use Exercise 5.27 to prove that Gal(E

,k) is solvable.
Solution. Absent.
(iv) Prove that Gal(E

,E)Gal(E

,k) and Gal(E

,k), Gal(E

,E)

=
Gal(E,k). Conclude that Gal(E,k) is solvable.
Solution. Absent.
5.29 Let f (x) Q[x] be an irreducible cubic with Galois group G.
(i) Prove that if f (x) has exactly one real root, then G

= S
3
.
Solution. If is a real root of f (x), then Q() R, and so it
cannot be the splitting eld E of f (x). By Theorem 4.31,
|G| = | Gal(E,Q)| = [E : Q] = [E : Q()][Q() : Q] 6.
Since G is isomorphic to a subgroup of S
3
, we must have G

= S
3
.
151
(ii) Find the Galois group of f (x) = x
3
2 Q[x].
Solution. f (x) is irreducible because it has no rational root. Its
discriminant is D = 108, and so its Galois group is S
3
.
(iii) Find a cubic polynomial g(x) Q[x] whose Galois group has
order 3.
Solution. Try g(x) = 3x
3
3x +1.
5.30 (i) If k is a eld and f (x) k[x] has derivative f

(x), prove that
either f

(x) = 0 or deg( f

) - deg( f ).
Solution. If f (x) = a
n
x
n
+ a
n1
x
n1
+ , then f (x) =
na
n
x
n1
+ (n 1)a
n1
x
n2
+ + a
1
. Hence, if f

(x) = 0,
then deg( f

) - n.
(ii) If k is a eld of characteristic 0, prove that an irreducible polyno-
mial p(x) k[x] has no repeated roots; that is, if E is the splitting
eld of p(x), then there is no a E with (x a)
2
| p(x) in E[x].
Solution. We may assume that p(x) is monic. By Exercise 3.67,
p(x) has no repeated roots if and only if gcd( p. p

) = 1, where
p

(x) is the derivative of p(x). But n = 0 in k because k has char-

acteristic 0, so that p

(x) = 0 and deg( p

) - deg( p). Since p(x)

is irreducible, its only monic divisors are 1 and itself; therefore,
gcd( p. p

) = 1 and p(x) has no repeated roots.

5.31 Let k be a eld of characteristic p.
(i) Prove that if f (x) =

i
a
i
x
i
k[x], then f

(x) = 0 if and only
if the only nonzero coefcients are those a
i
with p | i .
Solution. If f

(x) = 0, then any nonzero term a
i
x
i
of f (x)
gives the term i a
i
x
i 1
; if p a
i
, however, then i a
i
= 0, and
so f

(x) = 0. Conversely, if f (x) =

j
b
j
x
pj
, then f

(x) =

j
pj b
j
x
pj 1
= 0.
(ii) If k is nite and f (x) =

i
a
i
x
i
k[x], prove that f

(x) = 0 if
and only if there is g(x) k[x] with f (x) = g(x)
p
.
Solution. By part (i), if f

(x) = 0, then f (x) =

j
a
j p
x
j p
. For
each j , let b
p
j
= a
j p
. If g(x) =

j
b
j
x
j
, then
g(x)
p
= (

j
b
j
x
j
)
p
=

j
b
p
j
x
pj
= f (x).
(iii) Prove that if k is a nite eld, then every irreducible polynomial
p(x) k[x] has no repeated roots.
152
Solution. By Exercise 3.67, it sufces to prove that p

(x) = 0. But if
p

(x) = 0, then p(x) = g(x)

p
for some g(x) k[x], by part (ii), contra-
dicting p(x) being irreducible.
5.32 (i) If k = F
p
(t ), the eld of rational functions over F
p
, prove that
x
p
t k[t ] has repeated roots. (It can be shown that x
p
t is
an irreducible polynomial.)
Solution. Let be a root of x
p
t (in a splitting eld). Then
t =
p
and
x
p
t = x
p

p
= (x )
p
.
If p(x) is the irreducible polynomial of , then p(x) | (x )
p
,
and so it has repeated roots. (One can show that t
p
x is, in fact,
irreducible in k[t ].)
(ii) Prove that E = k() is a splitting eld of x
p
t over k.
Solution. Absent.
(iii) Prove that Gal(E,k) = {1}.
Solution. Absent.