Solutions to

TOPICS IN ALGEBRA
I.N. HERSTEIN

Part II: Group Theory

No rights reserved.

Any part of this work can be reproduced or transmitted in any form or by any
means.

Version: 1.1
Release: Jan 2013
Author: Rakesh Balhara
Preface

These solutions are meant to facilitate the deeper understanding of the book,
Topics in Algebra, 2nd edition. We have tried to stick with the notations devel-
oped in the book as far as possible. But some notations are extremely ambigu-
ous, so to avoid confusion, we resorted to alternate commonly used notations.

The following notation changes will be found in the text:

1. a mapping T operating on an element x is represented through T (x) rather

than xT .
2. subgroup generated by a is represented through hai rather than (a)
Any suggestions or errors are invited and can be mailed to: [email protected]

3
Problems (Page 35)

1. In the following determine whether the systems described are groups. If they
are not, point out which of the group axioms fail to hold.
(a) G = set of all integers, a · b ≡ a − b.
(b) G = set of all positive integers, a · b = ab, usual product of integers.
(c) G = a0 , a1 , · · · , a6 where

ai · aj = ai+j if i + j < 7,
ai · aj = ai+j if i + j ≥ 7

(for instance, a5 · a4 = a5+4−7 = a2 since 5 + 4 = 9 > 7).

(d) G = set of all rational numbers with odd denominators, a · b ≡ a + b, the
usual addition of rational numbers.
Solution:
(a) Clearly the binary operation is well-defined. Also a − b ∈ G ∀a, b ∈ G. So
the closure property also holds good. Now a · (b · c) = a · (b − c) = a − (b − c) =
a − b + c. On the other hand (a · b) · c = (a − b) · c = (a − b) − c = a − b − c. So
a · (b · c) 6= (a · b) · c. So the associativity does not hold good. Hence G is not a
group.

(b) Clearly the binary operation is well-defined. Also we can check the binary
operation satisfies closure and associativity too. Again 1 is the required identity
as 1 · a = a · 1 = a for all positive integers a. But we can see inverse of any
element except for 1 does not exist. So the existence of inverses fails. Hence G
is not a group.

(c) We can easily check G is a group with a0 as identity element and a7−i as
inverse element of ai .

(d) Again we can easily check G is a group with 0 as identity and − m

n as inverse
element of m
n .

2. Prove that if G is an abelian group, then for all a, b ∈ G and all integers n,
(a · b)n = an · bn .
Solution: We resort to induction to prove that the result holds for positive
integers. For n = 1, we have (a · b)1 = a · b = a1 · b1 . So the result is valid for
the base case. Suppose result holds for n = k − 1, i.e. (a · b)k−1 = ak−1 · bk−1 .
We need to show result also holds good for n = k. We have

(a · b)k = (a · b)k−1 · (a · b)
= (ak−1 · bk−1 ) · (a · b)
= (ak−1 · bk−1 ) · (b · a)
= (ak−1 · bk ) · a
= a · (ak−1 · bk )
= a k · bk

So the result holds for n = k too. Therefore, result holds for all n ∈ N. Next
suppose n ∈ Z. If n = 0, then (a·b)0 = e where e the identity element. Therefore
(a · b)0 = e = e · e = a0 · b0 . So the result is valid for n = 0 too. Next suppose n
is a negative integer. So n = −m, where m is some positive integer. We have

(a · b)n = (a · b)−m
= ((a · b)−1 )m by definition of the notation
= (b−1 · a−1 )m
= ((a−1 ) · (b−1 ))m
= (a−1 )m · (b−1 )m as the result is valid for positive integers
= (a−m ) · (b−m )
= an · bn

So the result is valid for negative integers too. Hence the result that (a · b)n =
an · bn holds in an abelian group for all n ∈ Z.

3. If G is a group such that (a · b)2 = a2 · b2 for all a, b ∈ G, show that G must

be abelian.
Solution: We have for all a, b ∈ G

(a · b)2 = a2 · b2
⇒ (a · b) · (a · b) = a2 · b2
⇒ a · ((b · a) · b) = a · ((a · b) · b)
⇒ (b · a) · b = (a · b) · b as a−1 exists in G
⇒ b · a = a · b as b−1 exists in G

But b · a = a · b ∀ a, b ∈ G implies G is an abelian group. Hence the result.

4. If G is a group in which (a · b)i = ai · bi for three consecutive integers i for

all a, b ∈ G, show that G is abelian.
Solution: Let n, n + 1, n + 2 be some three consecutive integers. Therefore we
have

(a · b)n = an · bn (1)
n+1 n+1 n+1
(a · b) =a ·b (2)
n+2 n+2 n+2
(a · b) =a ·b (3)

Using (2) we have

(a · b)n+1 = an+1 · bn+1

⇒ (a · b)n · (a · b) = an+1 · (bn · b)
⇒ (an · bn ) · (a · b) = (an+1 · bn ) · b, Using (1)
⇒ ((an · bn ) · a) · b = (an+1 · bn ) · b
⇒ (an · bn ) · a = (an · a) · bn
⇒ an · (bn · a) = an · (a · bn )
⇒ bn · a = a · bn (4)

Again using (3), analogously we have

bn+1 · a = a · bn+1
⇒ b · (bn · a) = a · bn+1
⇒ b · (a · bn ) = a · bn+1 , Using (4)
⇒ (b · a) · bn = (a · b) · bn
⇒b·a=a·b

So we have a · b = b · a ∀ a, b ∈ G. And hence G is abelian.

5. Show that the conclusion of the Problem 4 does not follow if we assume the
relation (a · b)i = ai · bi for just two consecutive integers.
Solution: Suppose (a · b)i = ai · bi for i = n and i = n + 1. We claim G
is abelian if and only if (a · b)n+2 = an+2 · bn+2 . Clearly, from last Problem
we have (a · b)n+2 = an+2 · bn+2 ⇒ G is abelian. Also if G is abelian, then
(a · b)i = ai · bi ∀ i ∈ Z; in particular result holds for i = n + 2. Thus G is
abelian if and only if (a · b)n+2 = an+2 · bn+2 . So the result of Problem 4 might
not follow if we assume (a · b)i = ai · bi for just two consecutive integers.

6. In S3 give an example of two elements x, y such that (x · y)2 6= x2 · y 2 .

Solution: We assume, as described in Example 2.2.3, S3 = {e, ψ, ψ 2 , φ, φ · ψ,
ψ · φ} with ψ 3 = e, φ2 = e and φ · ψ i · φ = ψ −i . We choose x = ψ and y = φ.
We have

(ψ · φ)2 = (ψ · φ) · (ψ · φ) = ψ · (φ · ψ · φ) = ψ · ψ −1 = e
Whereas
ψ 2 · φ2 = ψ 2 · e = ψ 2
Thus (ψ · φ)2 6= ψ 2 · φ2 .

7. In S3 show that there are four elements satisfying x2 = e and three elements
satisfying y 3 = e.
Solution: Again, as in Problem 6, we assume S3 = {e, ψ, ψ 2 , φ, φ · ψ, ψ · φ} with
ψ 3 = e, φ2 = e. We have (e)2 = e; (ψ)2 = ψ 2 ; (ψ 2 )2 = ψ; (φ)2 = e; (φ · φ)2 = e;
(φ · φ)2 = e. Thus e, φ, ψ · ψ, ψ · φ are the elements with their square equal to
identity. Also we have (e)3 = e; (ψ)3 = e; (ψ 2 )3 = e; (φ)3 = φ; (φ · ψ)3 = φ · ψ;
(ψ · φ)3 = ψ · φ. Thus we have e, ψ, ψ 2 with their square equal to identity. Hence
the result.

8. If G is a finite group, show that there exists a positive integer N such that
aN = e for all a ∈ G.
Solution: Since G is finite, we assume G = {g1 , g1 , · · · , gm } for some positive
integer m. For some gi ∈ G, consider the sequence gi , gi2 , gi3 , · · · . Since G is
finite and closed under binary operation, so there must be repetition in the se-
quence, i.e. gij = gik for some positive integers j and k with j > k. But that
means gij−k = e, where e is the identity element. Let j − k = ni . Thus we have
ni corresponding to every gi such that gini = e. Let N = n1 × n2 × · · · × nm . But
then giN = e for all i, showing the existence of required positive integer N .

9. (a) If the group G has three elements, show it must be abelian.

(b) Do part (a) if G has four elements.
(c) Do part (a) if G has five elements.
Solution:
(a) Let G be a group of order 3 and some a, b ∈ G with a 6= b.
Case 1 , Either of a or b equals to the identity element: Suppose a = e then
a · b = e · b = b = b · e = b · a. Similarly, if b = e, we have a · b = b · a. Thus if
either a or b equals to e, we have a · b = b · a.
Case 2 , Neither of a or b is identity element: Consider a · b. We have a · b 6= a,
otherwise it would mean b = e. Similarly a · b 6= b as a 6= e. Also G has only
three elements, so a · b has not option but to be equal to the identity element.
Therefore a · b = e. A similar argument will show that b · a = e. Thus a · b = b · a
in this case too.
So we have a · b = b · a ∀ a, b ∈ G. Hence G is abelian for o(G) = 3.

(b) Again let G be the group of order 4 and let some a, b ∈ G.

Case 1 , Either of a, b equals to e: In this case, clearly a · b = b · a.
Case 2 , Neither of a, b equals to e. Consider a · b. Clearly, a · b 6= a and a · b 6= b.
But since G has four elements, let c 6= e be the fourth element. So a · b has two
options, either equals to e or equals to c.
• If a · b = e, then a = b−1 ⇒ b · a = b · b−1 ⇒ b · a = e. Thus a · b = b · a = e.
• If a · b = c, then consider b · a. Clearly, b · a 6= a and b · a 6= b. So b · a has
only two options, either b · a = e or b · a = c. But if b · a = e, then it would
imply a · b = e, which is not true. So b · a = c too. Thus a · b = b · a = c.
Thus we have a · b = b · a for all a, b ∈ G. Hence G is abelian for o(G) = 4.

(c) For this part, we have to make use of the material not presented till now
in the book. Since the order of G is prime, therefore it is cyclic. But a cyclic
group is abelian, so G must be abelian for order 5.

10. Show that if every element of the group G has its own inverse, then G is
abelian.
Solution: Let some a, b ∈ G. So we have a−1 = a and b−1 = b. Also a · b ∈ G,
therefore a · b = (a · b)−1 = b−1 · a−1 = b · a. So we have a · b = b · a, showing G
is abelian.

11. If G is a group of even order, prove it has an element a 6= e satisfying a2 = e.

Solution: We prove the result by contradiction. Note that G is a finite group.
Suppose there is no element x satisfying x2 = e except for x = e. Thus if some
g 6= e belongs to G, then g 2 6= e, i.e. g 6= g −1 . It means every non-identity
element g has another element g −1 associated with it. So the non-identity ele-
ments can be paired into mutually disjoint subsets of order 2. We can assume
the count of these subsets equals to some positive integer n as G is a finite
group. But then counting the number of elements of G, we have o(G) = 2n + 1,
where 1 is added for the identity element. So G is a group of odd order, which
is not true. Hence there must exist an element a 6= e such that a2 = e for G is
a group of even order.

12. Let G be a nonempty set closed under the an associative product, which in
addition satisfies:
(a) There exists an e ∈ G such that a · e = a for all a ∈ G.
(b) Give a ∈ G, there exists an element y(a) ∈ G such that a · y(a) = e.
Prove that G must be a group under this product.
Solution: In order to show G is a group, we need to show
1. e · a = a ∀ a ∈ G, and

2. If a · y(a) = e, then y(a) · a = e.

Suppose some a ∈ G, therefore there exists y(a) ∈ G such that a · y(a) = e.
Again y(a) ∈ G implies there exists y(y(a)) ∈ G such that y(a) · y(y(a)) = e.
So we have

y(a) · a = (y(a) · a) · e
= (y(a) · a) · (y(a) · y(y(a)))
= ((y(a) · a) · y(a)) · y(y(a))
= (y(a) · (a · y(a)) · y(y(a))
= (y(a) · e) · y(y(a))
= y(a) · y(y(a))
=e

Thus
y(a) · a = e (1)
Now using (1), we have e · a = (a · y(a)) · a = a · (y(a) · a) = a · e = a. Thus

e·a=a (2)

Form (1) and (2), we conclude G is a group.

13. Prove, by an example, that the conclusion of Problem 12 is false if we as-

sume instead:
(a0 ) There exists an e ∈ G such that a · e = a for all a ∈ G.
(b0 ) Given a ∈ G, there exists
 y(a)∈ G such that  y(a) · a = e
a a +
Solution: Consider G = | a, b ∈ Q with usual matrix multiplica-
b b
+
tion as binary  operation
 ‘·’, where Q  denotes
 the positive rational numbers.
1 1 a a 1
We define e = . Also for M = , we define y(M ) = a+b e. One
0 0 b b
can easily see that M · e = M and y(M ) · M = e. Also G is not a group as
inverses do not exist.

14. Suppose a finite set G is closed under an associative product and that both
cancellation laws hold in G. Prove that G must be a group.
Solution: Since G is a finite group, so we can assume G = {g1 , g2 , . . . , gn }, for
some positive integer n. Let some a ∈ G. Consider S = {a · g1 , a · g2 , · · · , a · gn }.
We assert each element of S is distinct as if a · gi = a · gj with gi 6= gj , then
left-cancellation implies gi = gj . Therefore, o(S) = o(G), combining with the
fact that S ⊂ G, we conclude S = G. So if a ∈ G, therefore a ∈ S. But that
means a = a · gk , for some gk ∈ G. We claim gk is the right-identity. For
establishing our claim, we consider S 0 = {g1 · a, g2 · a, · · · , gn · a}. Again since
right-cancellation too holds good, proceeding in an analogous way, we can see
S 0 = G. Let some x ∈ G. So x ∈ S 0 , therefore x = gi · a for some gi ∈ G.
Now x · gk = (gi · a) · gk = gi · (a · gk ) = gi · a = x. Thus we have shown,
x · gk = x ∀ x ∈ G. So gk is the right-identity. Now, since gk ∈ G, therefore
gk ∈ S. So gk = a · gl , for some gl ∈ G. But that shows the existence of right-
inverse gl , for an arbitrarily chosen element a ∈ G. Thus right-inverse exists for
each element in G. With the existence of right-identity and right-inverses, we
concluded that G is group.

15. (a) Using the result of Problem 14, prove that the nonzero integers modulo
p, p a prime number, form a group under multiplication mod p.
(b) Do part (a) for the nonzero integers relatively prime to n under multiplica-
tion mod n.
Solution:
(a) Let G be the set consists of non-zero integers modulo p. We noticed that
G is a finite set with multiplication mod p as well-defined binary operation.
Using Problem 14, G would be a group if we show that the multiplication mod
p is associative and the both cancellation laws hold good. One can easily see
that a ⊗ (b ⊗ c) = (a ⊗ b) ⊗ c = abc mod p. So associativity is not an issue.
Next suppose a ⊗ b = a ⊗ c, we need to show that it would imply b = c. But
a ⊗ b = a ⊗ c ⇒ ab = ac mod p ⇒ a(b − c) = 0 mod p ⇒ p | a(b − c). But p
being prime, so either p | a or p | b − c. Since p6 | a, so p | (b − c). Also p6 | b and
p6 | c, so p | (b − c) implies b − c = 0, or b = c. Thus left-cancellation law holds
good. Similarly we can see right-cancellation also holds good. Using previous
problem, we conclude G is a group.
(b) Let G be the set consists of non-zero integers relatively prime to n. Clearly
G is a finite set. Also multiplication mod n is well-defined binary operation
over G. To show G is a group under multiplication mod n, we need to show
that associativity and both cancellation laws hold good. It is easy to see that
a ⊗ (b ⊗ c) = (a ⊗ b) ⊗ c = abc mod n. So associativity holds good. Next suppose
a⊗b = a⊗c. But that means ab = ac mod n ⇒ a(b−c) = 0 mod n ⇒ n | a(b−c).
But gcd(a, n) = 1, therefore n | a(b−c) implies n | (b−c). Also gcd(b, n) = gcd(c,
n) = 1, therefore n | (b − c) implies b = c. Thus we see left-cancellation holds
good. Similarly, we can check right-cancellation too holds good. And so G is a
group.

16. In Problem 14 show by an example that if one just assumed one of the
cancellation laws, then the conclusion need not follow.
Solution: We construct a set G with n ≥ 2 elements, equipped with a binary
operation · : G × G −→ G such that x · y = y. Clearly the binary operation · is
well-defined. Next, we see x · (y · z) = y · z = z. Also (x · y) · z = z. Thus the
binary operation is associative. Next we check left-cancellation. Suppose we
have x · y = x · z. But x · y = y and x · z = z. Thus x · y = x · z ⇒ y = z, showing
left-cancellation holds good. On the other hand, if we have x · y = z · y, we
cannot conclude x = z as x · y = z · y = y ∀x, z ∈ G. Thus right-cancellation
does not hold good. Finally, we prove G is not a group. Suppose G is group,
therefore it must have the identity element. Let e be the identity element. But
then x · e = e ∀x ∈ G, showing all elements are identity elements. Thus we get
G = {e}, but we have assumed G has n elements with n ≥ 2, hence a contra-
diction. Thus G has no identity element. Hence G is not a group. So if a finite
set with well-defined binary operation is satisfying associativity and one sided
cancellation law, it need not to be a group under that binary operation.

17. Prove that in Problem 14 infinite examples exist, satisfying the conditions,
which are not groups.
Solution: We define An = {in | i ∈ Z+ }, where n is a positive integer greater
than 1. Clearly An with usual multiplication as binary operation satisfies all
conditions of Problem 14, but is not a group as inverses do not exist for all ele-
ments. Also since there are infinite positive integers greater than 1, so we have
infinite examples satisfying the conditions of Problem 14 but are not groups.

18. For any n > 2, construct a non-abelian group of order 2n. (Hint: imitate
the relation in S3 )
Solution: Let G be the group that we are going to construct. Let · denotes its
binary operation and let e be its identity element. Thus we have constructed
an element of G, which is e. Next we construct an element a 6= e with order
n ≥ 3. Thus we have constructed n element, which are e, a, a2 , · · · , an−1 . Fi-
nally we construct an element other than already constructed, b, with order 2,
i.e. b2 = e. We interconnect a, b with the rule: b · a · b−1 = a−1 . We claim
we have got 2n elements, i.e. G = {e, a, a2 , · · · , an−1 , b, b · b, · · · , b · an−1 }. To
establish our claim, we first notice that b · ai · b = a−i . With this rule at hand,
one can easily check any expression resulting from the product of ai and bj will
belongs to those 2n elements. To give readers more insight, suppose we have
some expression ai · bj · ak . Now either j = 0 or j = 1. If j = 0, then expression
equals to ai+k , and thus belong to G. Whereas if j = 1, then let x = ai · b · ak .
We have b · x = (b · ai · b) · ak = a−i · ak = ak−i . Left-multiplying by b, we get
x = b · ak−i . Thus x ∈ G. Also a · b = b · a−1 ; since n 6= 2, therefore a · b 6= b · a.
Thus we have got G, a non-abelian group of order 2n.

19. If S is a set closed under an associative operation, prove that no matter how
you bracket a1 a2 · · · an , retaining the order of the elements, you get the same
element in S (e.g., (a1 · a2 ) · (a3 · a4 ) = a1 · (a2 · (a3 · a4 )); use induction on n)
Solution: Let Xn denote a expression we get by bracketing the n elements,
keeping the order of elements same. We need to show all expressions Xn are
same for all n. We will use induction over n. For n = 1 and n = 2, we have only
one expression possible, so trivially all expressions are same. For n = 3, we have
two ways of bracketing a1 a2 a3 , i.e a1 · (a2 · a3 ) and (a1 · a2 ) · a3 . Since the binary
operation · is satisfying associativity, therefore both expression are same. Next
suppose the result is true for n ≤ i−1. We need to show that the result is equally
true for n = i. Let Xi be some expression. Then it must be product of some
two shorter expressions, i.e Xi = Yα · Zi−α , where α < i. But all expressions
 0 0
with number of elements less than i are same. So Yα = a1 · Yα−1 , where Yα−1
0
is some expression consists of α − 1 elements. Similarly, Zi−α = aα+1 · Zi−α−1 .
Thus we have

Xi = Yα · Zi−α
0 0
= (a1 · Yα−1 ) · (aα+1 · Zi−α−1 )
0
= a1 · Xi−1 ,

where Xi−1 0 is some expression of i − 1 terms. But all expressions containing

i − 1 terms are same (order of elements is assumed to remain same). But then
0
all expression having i elements turns out to be equal to a1 · Xi−1 . Thus all
expressions having i elements are same. The induction hypothesis implies result
is valid for all n. Hence the result.

 
a b
20. Let G be the set of all real 2 × 2 matrices , where ad − bc 6= 0 is a
c d
rational number. Prove that the G forms a group under matrix multiplication.
Solution:
 It
 is easy to check that G is a group
 undermatrix multiplication,
1 0 1 d −b
with as identity element and (ad−bc) as inverse element of
 0 1 −c a
a b
. Note that ad − bc 6= 0 is given and of the inverses exist for all ele-
c d
ments.

 
a b
21. Let G be the set of all real 2 × 2 matrices where ad 6= 0. Prove
0 d
that G forms a group under matrix multiplication. Is G abelian?  0
a b0
 
a b
Solution: We have G closed under multiplication as =
0 d 0 d0
 0
aa ab0 + bd0

∈ G. Now since ad 6= 0, therefore G forms a group under ma-
0 dd0
   
1 0 1/a −b/ad
trix multiplication with as identity element and as in-
0 1 0 1/d
 0
a b0 a0 b0
       
a b a b a b
verse element of . Finally we have =
0 d 0 d 0 d0 0 d0 0 d
0 0 0 0 0 0 0 0
implies ab + bd = a b + b d. Since ab + bd 6= a b + b d for all values of a, b, d,
a0 , b0 , d0 , so G is not an abelian.

 
a 0
22. Let G be the set of all real 2 × 2 matrices where a 6= 0. Prove
0 a−1
that G is an abelian group under matrix multiplication.
 Solution: Easy to check G is a group under multiplication. Also we have
  0   0    0 
a 0 a 0 a 0 a 0 aa 0
= =
0 a−1 0 a0−1 0 a0−1 0 a−1 0 a−1 a0−1

Hence G is abelian too.

23. Construct in the G of Problem

 21 a subgroup
 of order 4.
1 0 −1 0
Solution: Let a = ,b= . Clearly a2 = I and b2 = I, where
0 −1 0 1
   
1 0 −1 0
I = . Also ab = ba = . Thus H = {I, a, b, ab} forms a
0 1 0 −1
subgroup of G.

 
a b
24. Let G be the set of all 2 × 2 matrices where a, b, c, d are integers
c d
modulo 2, such that ad − bc 6= 0. Using matrix multiplication as the operation
in G, prove that G is a group of order 6.
Solution: Its trivial to check that
           
1 0 0 1 0 1 1 0 1 1 1 1
G= , , , , ,
0 1 1 0 1 1 1 1 0 1 1 0

Hence o(G) = 6

 
a b
25. (a) Let G be the group of all 2×2 matrices where ad−bc 6= 0 and a,
c d
b, c, d are integers modulo 3, relative to matrix multiplication. Show o(G) = 48.
(b) If we modify the example of G in part (a) by insisting that ad − bc = 1, then
what is o(G)?
Solution: See Problem 26, which is general case of this problem.

 
∗ a b
26. (a) Let G be the group of all 2 × 2 matrices where a, b, c, d are the
c d
integers modulo p, p being a prime number, such that ad − bc 6= 0. G forms
group relative to matrix multiplication. What is o(G)?
(b) Let H be the subgroup of the G of part (a) defined by
  
a b
H= ∈ G | ad − bc = 1 .
c d

What is o(H)?
Solution:
(a) We will first count the number of ways in which ad − bc = 0. We separate
this counting into two cases. In first case we count the ways when ad = bc = 0.
In second case, we will count the number of ways in which ad = bc 6= 0.
Case 1 : ad = bc = 0: When a = 0, we can choose d in p ways; and when d = 0,
we can choose a in p ways. But in this a = d = 0 has been counted twice.
Therefore there are 2p − 1 ways in which ad = 0. Similarly, bc = 0 in 2p − 1
ways. Thus there are (2p − 1)2 ways in which ad = bc = 0.
Case 2 : ad = bc 6= 0: Since ad 6= 0, therefore a 6= 0 and d 6= 0. Similarly,
b 6= 0 and c 6= 0. We chose some value of a 6= 0 in p − 1 ways; some value of
d 6= 0 in p − 1 ways; some value of b in p − 1 ways; then find the value of c with
these chosen values of a, d, b. Since ad = bc mod p has unique solution in c for
non-zero values of a, d, b; thus we get unique value of c. Thus ad = bc 6= 0 can
be chosen in (p − 1)3 ways.
Finally, the number ways of choosing a, b, c, d with ad − bc 6= 0 equals number
ways of choosing a, b, c, d without any restriction minus the number of ways of
choosing a, b, c, d with ad−bc = 0. Thus the number of ways of choosing a, b, c, d
with ad − bc 6= 0 is p4 − (p − 1)3 − (2p − 1)2 . On simplification, we get

o(G) = p4 − p3 − p2 + p

(b) We separate our counting of a, b, c, d for which ad − bc = 1 in three separate

cases.
Case 1 : ad = 0: This restrict bc = −1. The number ways of choosing a, d for
which ad = 0 is: 2p − 1. On the other hand, the number of ways of choosing
b, c for which bc = −1 is: p − 1. Thus when ad = 0, we can choose a, b, c, d in
(2p − 1)(p − 1)
Case 2 : bc = 0: Analogous to previous case, we get number of ways of choosing
a, b, c, d as (2p − 1)(p − 1).
Case 3 : ad 6= 0 and bc 6= 0: In this case we get number of ways of choosing
a, b, c, d as (p − 1)(p − 1)
Thus total number of ways of choosing a, b, c, d for which ad − bc = 1 is: 2(2p −
(p − 1)2 . On simplifying we get
P
1)(p − 1) +
p−2

o(G) = p3 − p
Problems (Page 46)
1. If H and K are subgroups of G, show that H ∩ K is a subgroup of G.
(Can you see that the same proof shows that the intersection of any number of
subgroups of G, finite of infinite, is again a subgroup of G?)
Solution: Let some x, y ∈ H ∩ K. Therefore x ∈ H and y ∈ H. But H being
a subgroup, so xy −1 ∈ H. Again x ∈ K and y ∈ K. K being subgroup implies
xy −1 ∈ K. But xy −1 ∈ H and xy −1 ∈ K implies xy −1 ∈ H ∩ K. Thus we have
for all x, y ∈ H ∩ K, xy −1 ∈ H ∩ K. Hence H ∩ K is a subgroup of G.
Problems (Page 70)
1. Are the following mappings automorphisms of their respective groups?
(a) G group of integers under addition, T : x → −x.
(b) G group of positive reals under multiplication, T : x → x2 .
(c) G cyclic group of order 12, T : x → x3 .
(d) G is the group S3 , T : x → x−1
Solution:
(a) Yes, as G is an abelian group.

(b) No, as mapping is not a homomorphism.

(c) No, as if G = hai, then o(a) = 12, while o(T (a)) = 4, showing T is not an
automorphism. Note that an isomorphism preserve the order of elements.

(d) No, as G is a non-abelian group, so the mapping T : x −→ x−1 fails to be a

homomorphism.

2. Let G be a group, H a subgroup of G, T an automorphism of G. Let

T (H) = {T (h) | h ∈ H}. Prove T (H) is a subgroup of G.
Solution: We need to show T (H) is a subgroup of G. Let x, y ∈ T (H),
therefore x = T (h1 ) and y = T (h2 ) for some h1 , h2 ∈ H. But then xy −1 =
T (h1 )(T (h2 ))−1 = T (h1 )T (h−1 −1
2 ) = T (h1 h2 ) = T (h3 ) for some h3 ∈ H as H is
a subgroup of G. So xy −1 ∈ T (H) ∀ x, y ∈ H. This shows T (H) is a subgroup
of G.

3. Let G be a group, T an automorphism of G, N a normal subgroup of G.

Prove that T (N ) is a normal subgroup of G.
Solution: We need to show T (N ) is a normal subgroup of G. Clearly by last
Problem, T (N ) is a subgroup of G. Let some g ∈ G, therefore g = T (g 0 ) for some
g 0 ∈ G as T is an onto mapping. But then gT (N )g −1 = T (g 0 )T (N )T (g 0 )−1 =
T (g 0 )T (N )T (g 0−1 ) = T (g 0 N g 0−1 ) = T (N ) as N is normal in G. Thus we have
gT (N )g −1 = T (N ) ∀ g ∈ G. Hence T (N ) is normal in G.

4. For G = S3 prove that G ≈ I (G).

Solution: We have, from Lemma 2.8.2

G
≈ I (G) (1)
Z
Also we claim for Sn with n ≥ 3, we have Z = {I}, where I is the identity map-
ping. We assume A = {x1 , x2 , · · · , xn } is the set of elements on which mappings
from Sn operate. Suppose some T ∈ Sn with T 6= I, therefore there is some xi
such that T (xi ) 6= xi . Let T (xi ) = xj , where i 6= j. Since n ≥ 3, therefore we
can have xk with k 6= i and k 6= j. Define T 0 : A −→ A such that T 0 (xi ) = xi ;
T 0 (xj ) = xk ; T 0 (xk ) = xj ; and T 0 (xl ) = xl for the rest of xl ∈ A. But then
T T 0 (xi ) = T (T 0 (xi )) = T (xi ) = xj ; and T 0 T (xi ) = T 0 (T (xi )) = T 0 (xj ) = xk ,
showing T T 0 6= T 0 T . So if some T 6= I, then T T 0 6= T 0 T for some T 0 ∈ Sn .
Thus T ∈ / Z ∀ T 6= I; implying Z = {I}. Finally, using (1), we have
Sn ≈ I (Sn ) ∀ n ≥ 3

5. For any group G prove that I (G) is a normal subgroup of A (G) (the group
A (G)/I (G) is called the group of outer automorphisms of G).
Solution: We have some change of notations, so we prefer to give detailed
solution of this problem. We define Tg : G −→ G such that Tg (x) = gxg −1
(Note that this is different from what is given in the book). Clearly Tg ∈ A (G)
for all g ∈ G. Next, we define I (G) = {Tg | g ∈ G}. Let some some Tg1 ,
Tg2 ∈ I (G). But then for all x ∈ G, we have Tg1 Tg2 (x) = Tg1 (Tg2 (x)) =
Tg1 (g2 xg2−1 ) = g1 (g2 xg2−1 )g1−1 = (g1 g2 )x(g1 g2 )−1 = Tg1 g2 (x). Thus Tg1 Tg2 =
Tg1 g2 . But Tg1 g2 ∈ I (G), therefore Tg1 Tg2 ∈ I (G). Again let some Tg ∈ I (G),
therefore for all x ∈ G, we have Tg (x) = gxg −1 . So we have, for all x ∈ G

Tg (g −1 xg) = g(g −1 xg)g −1

⇒ Tg (g −1 xg) = x
⇒ Tg−1 (Tg (g −1 xg)) = Tg−1 (x)
⇒ g −1 xg = Tg−1 (x)
⇒ Tg−1 (x) = (g −1 )x(g −1 )−1
⇒ Tg−1 (x) = Tg−1 (x)

Thus Tg−1 = Tg−1 . But Tg−1 ∈ I (G), therefore Tg−1 ∈ I (G). Hence I (G) is
a subgroup of A (G). Finally, suppose some T ∈ A (G) and some Tg ∈ I (G),
then we have for all x ∈ G

T Tg T −1 (x) = T (Tg (T −1 (x)))

= T (g(T −1 (x))g −1 )
= T (g)T (T −1 (x))T (g −1 )
= T (g)x(T (g))−1
= TT (g) (x)

Thus T Tg T −1 = TT (g) . But TT (g) ∈ I (G). Thus T Tg T −1 ∈ I (G) ∀ T ∈

A (G). Hence I (G) is a normal subgroup in A (G).

6. Let G be a group of order 4, G = {e, a, b, ab}, a2 = b2 = e, ab = ba.

Determine A (G).
Solution: The possible proper subgroups of G are {e, a}, {e, b}, {e, ab}. We aim
at finding a mapping T which is an automorphism. Since T is homomorphism,
therefore T (e) = e. Also once we have found T (a) and T (b), the value of T (ab)
get decided by itself as T (ab) = T (a)T (b). Now what could be the possible
values of T (a) so that T is an automorphism. Since o(a) = 2, so o(T (a)) must
be 2. The elements with order 2 are a, b, ab, so T (a) has three choices namely
a, b, ab. Once T (a) is decided, what could be the possible values for T (b). Again
order of b is 2, so the order of T (b) must be 2. Again we have three possible
candidates, a, b, ab, out of which one has already been fixed to T (a). So T (b)
has two choices. Thus we have 3 × 2 = 6 possible automorphisms. Thus

     
e a b ab e a b ab e a b ab
A (G) = , , ,
e a b ab e a ab b e b a ab
     
e a b ab e a b ab e a b ab
, ,
e b ab a e ab a b e ab b a

7. (a) A subgroup C of G is said to be a characteristic subgroup of G if T (C) ⊂ C

for all automorphisms T of G. Prove a characteristic subgroup of G must be a
normal subgroup of G.
(b) Prove that the converse of (a) is false.
Solution:
(a) We, for some g ∈ G, define Tg : G −→ G such that Tg (x) = gxg −1 . It is
routine to check Tg is an automorphism of G for all g ∈ G. But it is given that
T (C) ⊂ C for all automorphisms T . So Tg (C) ⊂ C ∀ g ∈ G. But that means
gCg −1 ⊂ C, or gcg −1 ∈ C ∀ g ∈ G & ∀ c ∈ C. Thus C is normal in G.

(b) We simply give an example to show that converse of part(a) need not be
2 2
true. ForG = {e, a, b, ab}
 with a = e, b = e and ab = ba; and C = {e, a};
e a b ab
and T = , we have C is a normal subgroup of G but T (C) * C.
e b a ab
Note that T defined above is an automorphism of G.

8. For any group G, prove that the commutator subgroup G0 is a characteristic

subgroup of G. (See Problem 5, Section 2.7)
Solution: Let U = {xyx−1 y −1 | x, y ∈ G}. So GQ
0
= hU i. Also if some u ∈ U ,
then u−1 ∈ U too. So if some a ∈ G0 , then a = i∈Λ xi yi x−1 −1
i yi , where Λ is
some finite index set. But then
Y
T (a) = T ( xi yi x−1 −1
i yi )
i∈Λ
Y
= T (xi yi x−1 −1
i yi )
i∈Λ
Y
= T (xi )T (yi )T (x−1 −1
i )T (yi )
i∈Λ
Y
= T (xi )T (yi )(T (xi ))−1 (T (yi ))−1
i∈Λ
Y
= x0i yi0 x0−1
i yi
0−1

i∈Λ

So T (a) ∈ G0 ∀ a ∈ G0 . Thus T (G0 ) ⊂ G0 . Hence G0 is a characteristic sub-

group of G.

9. If G is a group, N a normal subgroup of G, M a characteristic subgroup of

N , prove that M is a normal subgroup of G.
Solution: Let some g ∈ G. We define Tg : G −→ G such that Tg (x) = gxg −1 .
Clearly Tg is an automorphism of G. Also Tg (N ) = N as N is given to be
normal in G. Now consider Tg : N −→ N . Since Tg (N ) = N , one can eas-
ily see Tg is an automorphism of N too. But then Tg (M ) ⊂ M as M is
given to be a characteristic subgroup of N . So gM g −1 ⊂ M ∀g ∈ G, or
gmg −1 ∈ M ∀ g ∈ G & ∀m ∈ M . Hence M is normal in G.

10. Let G be a finite group, T an automorphism of G with the property that

T (x) = x for x ∈ G if and only if x = e. Prove that every g ∈ G can be
represented as g = x−1 T (x) for some x ∈ G.
Solution: First note that G is given to be a finite group. We define mapping
φ : G −→ G such that φ(x) = x−1 T (x). Clearly mapping so defined is well-
defined. Also

φ(a) = φ(b) ⇒ a−1 T (a) = b−1 T (b)

⇒ T (a)(T (b))−1 = ab−1
⇒ T (ab−1 ) = ab−1

But T (x) = x implies x = e, so φ(a) = φ(b) implies ab−1 = e, i.e. a = b. Thus

mapping φ is one-to-one. But since G is finite, therefore φ being one-to-one
implies φ is onto too. But onto implies that if some g ∈ G, then it has its
pre-image in G, i.e. g = x−1 T (x) for some x ∈ G. Hence every element g of G
can be represented as x−1 T (x) for some x ∈ G.
 11. Let G be a finite group, T an automorphism of G with the property that
T (x) = x if and only if x = e. Suppose further that T 2 = I. Prove that G must
be abelian.
Solution: Using previous problem, if some g ∈ G, then g = x−1 T (x) for some
x ∈ G. So we have T (g) = T (x−1 T (x)) = T (x−1 )T (T (x)) = (T (x))−1 T 2 (x) =
(T (x))−1 x = (x−1 T (x))−1 = g −1 . Thus T (g) = g −1 ∀ g ∈ G. Now let
a, b ∈ G. So we have
T (ab) = (ab)−1 = b−1 a−1 (1)
Also
T (ab) = T (a)T (b) = a−1 b−1 (2)
Using (1) and (2), we have

b−1 a−1 = a−1 b−1 ⇒ ab = ba

So we have ab = ba ∀ a, b ∈ G. Hence G is an abelian group.

∗
12. Let G be a finite group and suppose the automorphism T sends more than
three-quarters of the elements of G onto their inverses. Prove that T (x) = x−1
for all a ∈ G and that G is abelian.
Solution: Define a set H = {x ∈ G | T (x) = x−1 }. So we have o(H) > 43 n.
Let some h ∈ H. Consider the set Hh. Clearly o(Hh) > 43 n. Therefore
o(H ∩ Hh) > n2 . Let some x ∈ H ∩ Hh. Therefore x = h1 and x = h2 h, for
some h1 , h2 ∈ H. Also since x ∈ H ∩ Hh ⊂ H, therefore T (x) = x−1 . So

T (x) = x−1 = h−1

1 (1)
T (x) = x = (h2 h) = h−1 h−1
−1 −1
2 (2)
T (x) = T (h2 h) = T (h2 )T (h) = h−1
2 h
−1
(3)

Using (2) and (3), we get h−1 h2−1 = h2−1 h−1 , or hh2 = h2 h. Also h2 = xh−1 ,
so hh2 = h2 h ⇒ hxh−1 = xh−1 h ⇒ hx = xh. Thus we have

hx = xh ∀ x ∈ H ∩ Hh (4)

Now consider N (h), i.e. normalizer subgroup of h. The equation (4) implies
H ∩ Hh ⊂ N (h). So o(N (h)) > n2 . But N (h) being a subgroup of G, there-
fore o(N (h)) | o(G), forcing N (h) = G. But that means hg = gh ∀ g ∈ G.
So h ∈ Z ∀ h ∈ H, where Z is the center subgroup of G. So o(Z) > 43 n.
Again Z being a subgroup of G, so o(Z) | o(G), forcing Z = G. But Z = G
implies G is abelian. Also if G is abelian, then if some x, y ∈ H, we have
T (xy −1 ) = T (x)T (y −1 ) = x−1 (y −1 )−1 = x−1 y = (y −1 x)−1 = (xy −1 )−1 , show-
ing xy −1 too belongs to H. Hence with G abelian, H becomes a subgroup of
G. But then o(H) | o(G), therefore H = G as o(H) > 34 n. But H = G implies
 T (x) = x−1 ∀ x ∈ G, what we need to show.

13. In Problem 12, can you find an example of a finite group which is non-
abelian and which has an automorphism which maps exactly three-quarters of
the elements of G onto their inverses?
Solution:
 Consider the Quaterniongroup Q = {±1, ±i, ±j, ±k}. Let T =
1 −1 i −i j −j k −k
. We left it to the reader to check T is
1 −1 −i i −j j k −k
an automorphism of Q which transfer exactly 43 elements of Q into their in-
verses.

∗
14. Prove that every finite group having more than two elements has a non-
trivial automorphism.
Solution: Let G be some group with order greater than 2. Now either G is
non-abelian, or G is abelian. If G is non-abelian, then we have Z 6= G, where Z
is the center subgroup of G. So there is some g ∈ G such that g ∈ / Z. We define
for some g ∈/ Z, mapping T : G −→ G such that T (x) = gxg −1 . We claim T so
defined is an automorphism which is not equal to identity mapping. T is an au-
tomorphism of G is easy to check. Also if T = I, then T (x) = x ∀ x ∈ G. But
that means gxg −1 = x ⇒ gx = xg ∀ x ∈ G, i.e. g ∈ Z which is not the case.
So T 6= I. So for non-abelian groups, we have found a non-trivial automorphism.

When G is abelian, either x2 = e for all x ∈ G or there is some x0 ∈ G such

that x20 6= e. If there is some x0 ∈ G such that x20 6= e, then we claim that the
mapping T : G −→ G such that T (x) = x−1 is a non-trivial automorphism of
G. Since G is abelian, T surely is an automorphism. Also if T = I, then we
have x−1 = x ∀ x ∈ G, i.e x2 = e ∀ x ∈ G, which is not the case. So T is a
non-trivial automorphism of G.

On the other hand, if G is abelian, with x2 = e ∀ x ∈ G, then for some positive

integer m
G ≈ Z2 × Z2 × · · · × Z2
| {z }
m times

Note that here we have assumed G to be finite. But then there exist m inde-
Q symbols, a1 , a2 , · · · , am with o(ai ) = 2 ∀ i. By independent we mean
pendent
ai 6= j∈Λ aj for any index set Λ. So

G = ha1 i × ha2 i × · · · × ham i

Also o(G) ≥ 3 guarantees us the existence of atleast two such independent

symbols, say a1 and a2 . We define a mapping T : G −→ G such that T (a1 ) = a2 ,
T (a2 ) = a1 and for the rest of independent symbols ai , T (ai ) = ai . Once defined
 for independent symbols, we extend it for all elements of G using:
Y Y
T( aj ) = T (aj )
j∈Λ j∈Λ

Clearly extending T this way for all elements of G makes T a homomorphism.

Also we can check T is onto. Thus we have found an automorphism of G which
is not equal to an identity mapping. Hence every finite group with order greater
than 2 has a non-trivial automorphism.

∗
15. Let G be a group of order 2n. Suppose that half of the elements of G are
of order 2, and the other half form a subgroup H of order n. Prove that H is
of odd order and is an abelian subgroup of G.
Solution: Let K denotes the set of all elements with order 2. Therefore G =
H ∪ K. Also the index of H in G is 2, so H is a normal subgroup of G (See
problem 2, page 53). Let some k ∈ K, therefore k −1 = k. Also H being
normal in G implies kHk −1 = H. Let some h ∈ H, therefore h = kh1 k −1 for
some h1 ∈ H. So we have h = kh1 k −1 ⇒ hk = kh1 ; but hk ∈ K (Why), so
hk = (hk)−1 . Thus kh1 = hk = (hk)−1 = k −1 h−1 = kh−1 ⇒ h1 = h−1 . Thus
we have h = kh−1 k ∀ h ∈ H. Now let some x, y ∈ H, therefore x = kx−1 k,
and y = ky −1 k. So

xy = (kx−1 k)(ky −1 k) = kx−1 kky −1 k = kx−1 y −1 k (1)

Also
xy = k(xy)−1 k = ky −1 x−1 k (2)
Using (1) and (2), we have

kx−1 y −1 k = ky −1 x−1 k ⇒ x−1 y −1 = y −1 x−1

⇒ yx = xy

So xy = yx ∀x, y ∈ H. Hence H is an abelian subgroup of G Also since there

is no element with order 2, so order of H is odd (See Problem 11, page 35).

∗
16. Let φ(n) be the Euler φ-function. If a > 1 is an integer, prove that
n | φ(an − 1).
Solution: Let G be a cyclic group of order an − 1. The existence of such
G is guaranteed (Why). Therefore G = hxi for some x ∈ G. Define mapping
T : G −→ G such that T (y) = y a . Now since gcd(a, an −1) = 1, therefore T is an
2
automorphism of G, i.e T ∈ A (G). Also T 2 (y) = T T (y) = T (y a ) = y a . So we
n n
have T n (y) = y a . But y a −1 = e, where e is the identity element of G, therefore
n
y a = y. So T n (y) = y ∀ y ∈ G. Therefore T n = I, where I is the identity
mapping. Next suppose for some positive integer m < n, we have T m = I.
Then we have T m (y) = y ∀ y ∈ G, in particular, we have T m (x) = x. But
 m m
T m (x) = x ⇒ xa = x ⇒ xa −1 = e ⇒ o(x) < an − 1, which is not true. Thus
n is the smallest possible integer for which T n = I. Thus o(T ) = n in A (G).
Also G being finite cyclic group, therefore o(A (G)) = φ(o(G)) = φ(an − 1). So
by Lagrange Theorem, we get n | φ(an − 1).

17. Let G be a group and Z the center of G. If T is any automorphism of G,

prove that Z(T ) ⊂ Z.
Solution: We need to show that Z is a characteristic subgroup of G. For some
automorphism mapping T , we need to show T (Z) ⊂ Z. Let some z ∈ T (Z),
therefore there exist some x ∈ Z such that T (x) = z. But x ∈ Z im-
plies xg = gx ∀ g ∈ G. Since mapping T is one-to-one, therefore we have
T (xg) = T (gx) ∀ g ∈ G. But T (xg) = T (gx) ⇒ T (x)T (g) = T (g)T (x) ⇒
zT (g) = T (g)z ∀ g ∈ G. But since T is onto, therefore zT (g) = T (g)z ∀ g ∈
G ⇒ zg 0 = g 0 z ∀ g 0 ∈ G, which says z ∈ Z. Therefore z ∈ T (Z) ⇒ z ∈ Z. So
T (Z) ⊂ Z.

18. Let G be a group and T an automorphism of G. If, for a ∈ G, N (a) = {x ∈

G | xa = ax}, prove that N (T (a)) = T (N (a)).
Solution: We need to show N (T (a)) = T (N (a)) ∀ a ∈ G & T ∈ A (G). We
have N (T (a)) = {x ∈ G | xT (a) = T (a)x}. But since T is an onto mapping,
therefore

N (T (a)) = {T (x0 ) | x0 ∈ G & T (x0 )T (a) = T (a)T (x0 )}

= {T (x0 ) | x0 ∈ G & T (x0 a) = T (ax0 )}
= {T (x0 ) | x0 ∈ G & x0 a = ax0 }; since mapping is one-to-one
= {T (x0 ) | x0 ∈ N (a)}
= T (N (a))

19. Let G be a group and T an automorphism of G. If N is normal subgroup of

G such that T (N ) ⊂ N , show how you could use T to define an automorphism
of G/N .
Solution:
20. Use the discussion following Lemma 2.8.3 to construct
(a) a non-abelian group of order 55.
(b) a non-abelian group of order 203.
Solution:
21. Let G be the group of order 9 generated by elements a, b, where a3 = b3 = e.
Find all automorphisms of G.
Solution: Again, as in Problem 6, we are going to exploit the fact that an
automorphism leaves the order of elements unchanged. G has four non-trivial
subgroups, which are {e, a, a2 }; {e, b, b2 }; {e, a2 b, ab2 }; {e, ab, a2 b2 }. Now the
element e has not choice but has to map to itself, so only 1 choice. The element
a has 8 choices for being get mapped as all elements other than e has same
order as a has. Once image of a is fixed, the image of a2 get fixed by itself as
T (a2 ) = T (a)T (a). So a2 has no choice. Next after fixing image of e, a, a2 , the
element b has only 6 choices left. But fixing the image of a andb fixes the image
of remaining elements. So b2 , ab, a2 b2 , a2 b, ab2 have no choices. Thus the total
automorphisms of G are 8 × 6 = 48. So o(A (G)) = 48.
Problems (Page 74)
1. Let G be a group; consider the mappings of G into itself, λg , defined for
g ∈ G by λg (x) = xg for all x ∈ G. Prove that λg is one-to-one and onto, and
that λgh = λh λg .
Solution: We start with one-to-one. Suppose λg (x) = λg (y). But λg (x) =
λg (y) ⇒ xg = yg ⇒ x = y. So λg (x) = λg (y) implies x = y, showing λg is
one-to-one mapping. Next we tackle onto. Suppose some y ∈ G. But then for
x = yg −1 ∈ G, we have λg (x) = (yg −1 )g = y. This shows x is the inverse-
image of y. Thus λg is onto too. Finally, we have λgh (x) = xgh = (λg (x))h =
λh (λg (x)) = λh λg (x) for all x ∈ G. Thus λgh = λh λg .

2. Let λg be defined as in Problem 1, τg as in the proof of Theorem 2.9.1. Prove

that for any g, h ∈ G, the mappings λg , τh satisfy λg τh = τh λg . (Hint: For
x ∈ G consider λg τh (x) and τh λg (x).)
Solution: We recap our notation, mapping λg : x −→ xg and τh : x −→ hx.
Now we have for all x ∈ G, λg τh (x) = λg (τh (x)) = λg (hx) = hxg = τh (xg) =
τh (λg (x)) = τh λg (x). Thus λg τh = τh λg .

3.If θ is one-to-one mapping of G onto itself such that λg θ = θλg for all g ∈ G,
prove that θ = τh for some h ∈ G.
Solution: We are given θ is an one-to-one and onto mapping with λg θ =
θλg ∀ g ∈ G. So we have λg θ(x) = θλg (x) ∀ g, x ∈ G ⇒ θ(x)g = θ(xg) ∀ g,
x ∈ G. Since it holds for all x ∈ G, in particular must hold for x = e. Thus we
have θ(e)g = θ(g). Since θ(e) ∈ G, so we let θ(e) = h for some h ∈ G. Thus we
have θ(g) = hg ∀ g ∈ G. But that means θ = τh , hence the result.

4. (a) If H is a subgroup of G show that for every g ∈ G, gHg −1 is a subgroup

of G.
(b) Prove that W = intersection of all gHg −1 is a normal subgroup of G.
Solution:
(a) Consider gHg −1 for some g ∈ G. Let x, y ∈ gHg −1 , therefore x = gh1 g −1
and y = gh2 g −1 for some h1 , h2 ∈ H. But then xy −1 = gh1 g −1 (gh2 g −1 )−1 =
gh1 g −1 gh−1
2 g
−1
= gh1 h−1
2 g
−1
= gh3 g −1 for some h3 ∈ H. Thus xy −1 ∈
−1
gHg ∀ x, y ∈ gHg . Hence gHg −1 is a subgroup of G for all g ∈ G.
−1

gHg −1 . Let some x ∈ G. Consider xW x−1 . We

T
(b) We have W =
g∈G
have xW x−1 gHg −1 )x−1 = x(gHg −1 )x−1 = xgHg −1 x−1 =
T T T
= x(
g∈G g∈G g∈G
(xg)H(xg)−1 . But since the mapping φ : G −→ G such that φ(g) = xg is
T
g∈G
an onto mapping, so xW x−1 = (xg)H(xg)−1 =
T T 0 0−1
g Hg = W . But
g∈G g 0 ∈G
xW x−1 = W ∀ x ∈ G implies W is normal in G.

5. Using Lemma 2.9.1 prove that a group of order p2 , where p is a prime num-
ber, must have a normal subgroup of order p.
Solution: Let G be a group with order p2 . Since p | o(G) and p is a prime num-
ber, therefore the Cauchy’s Theorem guarantees us existence of an element a of
order p. Let H = hai. Therefore H is a subgroup of order p. Now since p2 6 | p!,
therefore we have o(G)6 | iG (H)!. So using Lemma 2.9.1, we have Kθ 6= {e},
along with Kθ ⊂ H and Kθ normal in G. But then o(H) = p, a prime number,
forces Kθ = H. Thus H is normal in G.

6. Show that in a group G of order p2 any normal subgroup of order p must lie
in the center of G.
Solution: Let H be the normal subgroup of G. Let some h ∈ H and some
g ∈ G. We have ghg −1 ∈ H as H is normal in G. If h = e, then we have
ghg −1 = h, or gh = hg ∀ g ∈ G. Next we assume h 6= e, so H = hhi. Now
since the o(g) | o(G), therefore o(g) = 1, p, p2 . When o(g) = 1, then we have
g = e, and so ghg −1 = g, or hg = hg. When o(g) = p, we have ghg −1 ∈ H.
Since H = hhi, therefore ghg −1 = hi for some positive integer i < p. With this
p
have h = g p hg −p = g p−1 (ghg −1 )g −(p−1) = g p−1 (hi )g −(p−1) = · · · = hi . So
p
hi −1 = e, but o(h) = p, therefore p | ip − 1, or ip = 1 mod p. But we have,
from Fermat theorem ip = i mod p. From these two equations, we conclude
i = 1 mod p, or i = 1. Thus we have ghg −1 = h, or gh = hg, for all h ∈ H and
for all g of order p. Finally, when o(g) = p2 , we have G = hgi, therefore G is
abelian and we have gh = hg in this case too. Thus we concluded gh = hg for
all h ∈ H and for all g ∈ G. And so h ∈ Z ∀ h ∈ H, where Z is the center
subgroup of G. Hence H ⊂ Z.

7. Using the result of Problem 6, prove that any group of order p2 is abelian.
Solution: Let G be a group with order p2 . Problem 5 implies that there exist
a normal subgroup H of order p. But since p is a prime number, therefore
H = hai for some a ∈ H. But since o(G) = p2 , therefore there exist b ∈ G
such that b ∈/ H. Let K = hbi. Now order of b divides order of G, therefore
o(b) = 1 or p or p2 . If o(b) = 1, then we have b = e ∈ H, which is contravene our
assumption. So o(b) 6= 1. Next if o(b) = p, then since o(G)6 | iG (K)!, we have
K too normal in G. Therefore, Problem 6 implies H, K ⊂ Z. And so HK ∈ Z.
But o(HK) = o(H)o(K)/o(H ∩ K) = p2 as H ∩ K = {e}. Therefore o(Z) ≥ p2 ,
forcing Z = G, making G an abelian group. Finally if o(b) = p2 , then we have
G = hbi and hence abelian in this case too. So we concluded G is an abelian
group.

8. If p is a prime number, prove that any group G of order 2p must have a

 subgroup of order p, and that this subgroup is normal in G.
Solution: Cauchy Theorem guarantees the existence of an element a ∈ G such
that o(a) = p. But then H = hai is a subgroup of G with o(H) = p. Also since
2p6 | 2!, therefore o(G)6 | iG (H)!. But then there exists a normal subgroup K of
G inside H with K 6= {e}. So we have o(K) | o(H) and with o(H) being prime
forces K = H. Thus H is normal in G.

9. If o(G) is pq where p and q are distinct prime numbers and if G has a normal
subgroup of order p and a normal subgroup of order q, prove that G is cyclic.
Solution: We are given a group G with o(G) = pq, where p, q are prime
numbers. Also we are given H, K normal subgroups of G with o(H) = p and
o(K) = q. So we have H = hai for some a ∈ H; and K = hbi for some
b ∈ K. Also since p and q are distinct primes, so we have H ∩ K = {e}.
We claim ab = ba. To establish our claim, consider aba−1 b−1 . We have
aba−1 b−1 = a(ba−1 b−1 ) = ah for some h ∈ H. Therefore aba−1 b−1 ∈ H. Again,
aba−1 b−1 = (aba−1 )b−1 = kb−1 for some k ∈ K. Therefore aba−1 b−1 ∈ K. So
aba−1 b−1 ∈ H ∩ K = {e}. That is aba−1 b−1 = e ⇒ ab = ba. Next we claim
o(ab) = pq, so that G = habi, i.e. a cyclic group. We have (ab)pq = apq bpq ,
as ab = ba. Therefore (ab)pq = apq bpq = (ap )q (bq )p = eq ep = e. Therefore
pq | o(ab). Suppose for some positive integer t with t < pq, we have (ab)t = e.
But (ab)t = e ⇒ at bt = e ⇒ at = b−t ⇒ atq = (bq )−t ⇒ atq = e ⇒ o(a) | tq ⇒
p | tq ⇒ p | t as gcd(p, q) = 1. Similarly, we have q | t. But then we have pq | t,
implying t > pq as p 6= 0, which is against our assumption. So we have pq as
the smallest positive integer for which (ab)pq = e, implying o(ab) = pq. Thus
G = habi and is cyclic.

∗
10. Let o(G) be pq, p > q are primes, prove
(a) G has a subgroup of order p and a subgroup of order q.
(b) If q6 | p − 1, then G is cyclic.
(c) Given two primes p, q, q | p − 1, there exists a non-abelian group of order pq.
(d) Any two non-abelian groups of order pq are isomorphic.
Solution:
(a)
Problems (Page 80)
1. Find the orbit and cycles of the following
 permutations:
1 2 3 4 5 6 7 8 9
(a)
2 3 4 5 1 6 7 9 8
 
1 2 3 4 5 6
(b)
6 5 4 3 1 2
Solution:  
1 2 3 4 5 6 7 8 9
(a) Clearly = (1, 2, 3, 4, 5)(6)(7)(8, 9). So orbit
2 3 4 5 1 6 7 9 8
of 1, 2, 3, 4 and 5 is the set {1, 2, 3, 4, 5}; orbit of 6 is 6; orbit of 7 is 7; orbit of
8 and 9 is the set {8, 9}. Also (1, 2, 3, 4, 5) and (8, 9) are its cycles.
 
1 2 3 4 5 6
(b) Again = (1, 6, 2, 5)(3, 4). So the orbit of 1, 2, 5 and 6
6 5 4 3 1 2
is the set {1, 2, 5, 6}; and the orbit of 3 and 4 is the set {3, 4}. Also (1, 6, 2, 5)
and (3, 4) are its cycles.

2. Write the permutation

 in the Problem 1 as the product
 of disjoint cycles.
1 2 3 4 5 6 7 8 9
Solution: We have = (1, 2, 3, 4, 5)(6)(7)(8, 9)
2 3 4 5 1 6 7 9 8
 
1 2 3 4 5 6
and = (1, 6, 2, 5)(3, 4).
6 5 4 3 1 2

3. Express as the product of disjoint cycles:

(a) (1, 5)(1, 6, 7, 8, 9)(4, 5)(1, 2, 3).
(b) (1, 2)(1, 2, 3)(1, 2).
Solution:
(a) Let (1, 5)(1, 6, 7, 8, 9)(4, 5)(1, 2, 3) = τ . So we have τ = τ1 τ2 τ3 τ4 , where
τ1 = (1, 5), τ2 = (1, 6, 7, 8, 9), τ3 = (4, 5) and τ4 = (1, 2, 3). Now

τ (1) = τ1 τ2 τ3 τ4 (1)
= τ1 (τ2 (τ3 (τ4 (1))))
= τ1 (τ2 (τ3 (2)))
= τ1 (τ2 (2))
= τ1 (2)
=2

Repeating analogously, we have τ (2) = 3; τ (3) = 6; τ(6) = 7; τ (7) = 8; τ (8) = 9; 

1 2 3 4 5 6 7 8 9
τ (9) = 5; τ (5) = 4; and τ (4) = 1. Thus we have τ = =
2 3 6 1 4 7 8 9 5
(1, 2, 3, 6, 7, 8, 9, 5, 4).

(b) Proceeding as in part (a), we have (1, 2)(1, 2, 3)(1, 2) = (1, 3, 2).
4. Prove that (1, 2, . . . , n)−1 = (n, n − 1, n − 2, . . . , 2, 1).
Solution: One can easily check (1, 2, . . . , n)(n, n − 1, . . . , 1) = I, where I is the
identity permutation. Hence (1, 2, . . . , n)−1 = (n, n − 1, . . . , 1).

5. Find the cycle structure of all the powers of (1, 2, . . . , 8).

Solution: Let (1, 2, 3, 4, 5, 6, 7, 8) = τ . So we have

τ 2 = τ τ = (1, 2, 3, 4, 5, 6, 7, 8)(1, 2, 3, 4, 5, 6, 7, 8)
= (1, 3, 5, 7)(2, 4, 6, 8)
τ = τ 2 τ = (1, 3, 5, 7)(2, 4, 6, 8)(1, 2, 3, 4, 5, 6, 7, 8, 9)
3

= (1, 4, 7, 2, 5, 8, 3, 6)
τ = τ 3 τ = (1, 4, 7, 2, 5, 8, 3, 6)(1, 2, 3, 4, 5, 6, 7, 8, 9)
4

= (1, 5)(2, 6)(3, 7)(4, 8)

τ = τ 4 τ = (1, 5)(2, 6)(3, 7)(4, 8)(1, 2, 3, 4, 5, 6, 7, 8, 9)
5

= (1, 6, 3, 8, 5, 2, 7, 4)
τ = τ 5 τ = (1, 6, 3, 8, 5, 2, 7, 4)(1, 2, 3, 4, 5, 6, 7, 8, 9)
6

= (1, 7, 5, 3)(2, 8, 6, 4)
τ = τ 6 τ = (1, 7, 5, 3)(2, 8, 6, 4)(1, 2, 3, 4, 5, 6, 7, 8, 9)
7

= (1, 8, 7, 6, 5, 4, 3, 2)
τ = τ 7 τ = (1, 8, 7, 6, 5, 4, 3, 2)(1, 2, 3, 4, 5, 6, 7, 8, 9)
8

= (1)(2)(3)(4)(5)(6)(7)(8) = I

So for i ∈ Z, we have τ i = τ i mod 8

 Solutions to

TOPICS IN ALGEBRA
I.N. HERSTEIN

Part III: Ring Theory

No rights reserved.

Any part of this work can be reproduced or transmitted in any form or by any
means.

Version: 1.1
Release: Jan 2013
Author: Rakesh Balhara
Preface

These solutions are meant to facilitate deeper understanding of the book, Topics
in Algebra, second edition, written by I.N. Herstein. We have tried to stick with
the notations developed in the book as far as possible. But some notations are
extremely ambiguous, so to avoid confusion, we resorted to alternate commonly
used notations.

The following notation changes will be found in the text:

1. use of unity element or simply unity instead of unit element.
2. use of unit element or simply unit in place of only unit.
3. an ideal generated by a is denoted by hai instead of (a).

4. use of gcd(a, b) instead of (a, b) for greatest common divisor of a, b.

Also following symbols are used in the text without any description, unless some
other symbol is specifically described in the problem statement for the same:
1. N is used for natural numbers, i.e. 1, 2, 3, · · · .

2. Z is used for integers, i.e. · · · , −2, −1, 0, 1, 2, · · · .

3. W is used for whole numbers, i.e. 0, 1, 2, · · · .
4. Zp is used for ring of integers with addition modulo p and multiplication
modulo p as its addition and multiplication respectively.

Any suggestions or errors are invited and can be mailed to: [email protected]

3
Problems (Page 130)
R is a ring in all the problems.

1. If a, b, c, d ∈ R, evaluate (a + b)(c + d).

Solution: We have
(a + b)(c + d) = a(c + d) + b(c + d)
= ac + ad + bc + bd
So (a + b)(c + d) = ac + ad + bc + bd.

2. Prove that if a, b ∈ R, then (a + b)2 = a2 + ab + ba + b2 , where by x2 we

mean xx.
Solution: We have
(a + b)2 = (a + b)(a + b)
= a(a + b) + b(a + b)
= aa + ab + ba + bb
= a2 + ab + ba + b2
Hence the result.

3. Find the form of the binomial theorem in a general ring; in other words, find
an expression for (a + b)n , where n is a positive integer.
Solution: We claim
X
(a + b)n = x1 x2 · · · xn
xi =a or b

We establish our claim Pby induction over n. For base case n = 1, we have
(a + b)1 = a + b = x1 . So for n = 1, expression is valid. Suppose the
x1 =a or b
expression (a + b)n =
P
x1 x2 · · · xn is valid for n = m − 1, we will show
xi =a or b
the expression is then valid for n = m too. We have
(a + b)m = (a + b)m−1 (a + b)
!
X
= x1 x2 · · · xm−1 (a + b)
xi =a or b
! !
X X
= x1 x2 · · · xm−1 a+ x1 x2 · · · xm−1 b
xi =a or b xi =a or b
! !
X X
= x1 x2 · · · xm−1 a + x1 x2 · · · xm−1 b
xi =a or b xi =a or b
 X
= x1 x2 · · · xm−1 xm
xi =a or b

Thus the expression is equally valid for n = m. So we have for all n ∈ N,

X
(a + b)n = x1 x2 · · · xn
xi =a or b

4. If every x ∈ R satisfies x2 = x, prove that R must be commutative. (A ring

in which x2 = x for all elements is called a Boolean ring.)
Solution: We are given x2 = x ∀ x ∈ R. So for all x, x2 = 0 ⇒ x = 0 as
x2 = x. But we have ∀ x, y ∈ R,

(xy − xyx)2 = (xy − xyx)(xy − xyx)

= xyxy − xyxyx − xyx2 y + xyx2 yx
= xyxy − xyxyx − xyxy + xyxyx Using x2 = x
=0

But
(xy − xyx)2 = 0 ⇒ xy − xyx = 0 (1)
Similarly, we can see (yx − xyx)2 = 0. Therefore

yx − xyx = 0 (2)

Using (1) and (2) we have xyx = xy = yx. So xy = yx ∀ x, y ∈ R. Hence R

is commutative.

5. If R is a ring, merely considering it as an abelian group under its addition,

we have defined, in Chapter 2, what is meant by na, where a ∈ R and n is an
integer. Prove that if a, b ∈ R and n, m are integers, then (na)(mb) = (nm)(ab).
Solution: We have

(na)(mb) = (a + · · · + a)(b + · · · + b)
| {z } | {z }
n times m times
= a(b + · · · + b) + · · · + a(b + · · · + b)
| {z } | {z }
m times m times
| {z }
n times
= (ab + · · · + ab) + · · · + (ab + · · · + ab)
| {z } | {z }
m times m times
| {z }
n times
= m(ab) + · · · + m(ab)
| {z }
n times
 = (nm)(ab)

Hence the result.

6. If D is an integral domain and D is of finite characteristic, prove that

characteristic of D is a prime number.
Solution: Let the characteristic of D be p, therefore pa = 0 ∀ x ∈ D and p is
the smallest such positive integer. Suppose p is not a prime, therefore p = rs for
some positive integers r and s, with both not equal to 1. Let some a 6= 0 ∈ D,
therefore a2 ∈ D too. So we have

pa2 = 0
⇒ (rs)(aa) = 0
⇒ (ra)(sa) = 0

D being an integral domain, implies ra = 0 or sa = 0. When ra = 0, we have

∀x∈D

(ra)x = 0
⇒ (a + a + · · · + a)x = 0
| {z }
r times
⇒ (ax + ax + · · · + ax) = 0
| {z }
r times
⇒ a(x + x + · · · + x) = 0
| {z }
r times
⇒ a(rx) = 0 (1)

But a 6= 0 and D an integral domain, therefore (1) implies rx = 0. So we have

rx = 0 ∀ x ∈ D with 1 < r < p, which is a contradiction as p is the smallest
such integer. Similarly, when sa = 0 we have contradiction. Thus p = rs is not
possible, thereby proving p is a prime.

7. Give an example of an integral domain which has an infinite number of ele-

ments, yet is of finite characteristic.
Solution: We define Zp [x] = {am xm + am−1 xm−1 + · · · a1 x + a0 | ai ∈ Zp , m ∈
W}, where Zp is a field of integers modulo p, p being prime. Clearly pf (x) =
0 ∀ f (x) ∈ Zp [x]. Also Zp [x] has infinite number of elements. So Zp [x] is the
desired example.

8. If D is an integral domain and if na = 0 for some a 6= 0 in D and some

integer n 6= 0, prove that D is of finite characteristic.
Solution: We are given na = 0 for some a ∈ D with a 6= 0 and n ∈ N. We
have ∀x ∈ D

(na)x = 0
⇒ (a + a + · · · + a)x = 0
| {z }
n times
⇒ (ax + ax + · · · + ax) = 0
| {z }
n times
⇒ a(x + x + · · · + x) = 0
| {z }
n times
⇒ a(nx) = 0 (1)

6 0 and D being integral domain, (1) implies nx = 0. So we have

With a =
nx = 0 ∀ x ∈ D, showing D is of finite characteristic.

9. If R is a system satisfying all the conditions for a ring with unit element with
possible exception of a + b = b + a, prove that the axiom a + b = b + a must
hold in R and that R is thus a ring. (Hint: Expand (a + b)(1 + 1) in two ways.)
Solution: We have for a, b ∈ R

(a + b)(1 + 1) = a(1 + 1) + b(1 + 1)

=a+a+b+b (1)

Also we have

(a + b)(1 + 1) = (a + b)1 + (a + b)1

=a+b+a+b (2)

From (1) and (2) we have a + a + b + b = a + b + a + b, or a + b = b + a. Thus

axiom a + b = b + a holds true in R, thereby proving R is a ring.

10. Show that the commutative ring D is an integral domain if and only if for
a, b, c ∈ D with a 6= 0 the relation ab = ac implies that b = c.
Solution: Suppose D is an integral domain. Now for a 6= 0, the relation

ab = ac ⇒ ab − ac = 0
⇒ a(b − c) = 0

But a 6= 0 and D an integral domain, imply b − c = 0, or b = c. Thus the

relation ab = ac with a 6= 0 implies b = c.

Conversely, suppose D is a commutative ring with a 6= 0 and ab = ac implying

b = c. Now suppose xy = 0 for some x, y ∈ D. If x 6= 0, then xy = 0 = x.0.
But xy = x0 with x 6= 0 implies y = 0. So xy = 0 and x 6= 0 implies y = 0.
Similarly, xy = 0 and y 6= 0 implies x = 0. Therefore xy = 0 implies x = 0 or
y = 0. Hence D is an integral domain.

11. Prove that Lemma 3.3.2 is false if we drop the assumption that the integral
domain is finite.
Solution: When D is infinite, Da = {da | d ∈ D} might not be equal to D
for some a ∈ D, the fact which we had used to prove the Lemma 3.3.2. For
example in the ring of integers Z, which is an infinite integral domain, 2Z 6= Z.
Also Z is not a field. Thus an infinite integral domain might not be a field.

12. Prove that any field is an integral domain.

Solution: Let F be some field and xy = 0 for some x, y ∈ F . If x 6= 0, then
there must exist x0 , multiplicative inverse of x in F . So we have

xy = 0 ⇒ x0 (xy) = x0 (0)
⇒ (x0 x)y = 0
⇒ (1)y = 0
⇒y=0

Similarly, when y 6= 0, we have x = 0. So xy = 0 implies x = 0 or y = 0.

Therefore F is an integral domain. Hence any field is an integral domain.

13. Useing the pigeonhole principle, prove that if m and n are relatively
prime integers and a and b are any integers, there exist an integer x such
that x ≡ a mod m and x ≡ b mod n. (Hint: Consider the remainders of
a, a + m, a + 2m, . . . , a + (n − 1)m on division by n.)
Solution: Consider the remainders of a, a + m, a + 2m, . . . , a + (n − 1)m on di-
vision by n. We claim no two remainder is same. Suppose if (a + im) mod n =
(a + jm) mod n, then (a + im) ≡ (a + jm) mod n ⇒ m(i − j) ≡ 0 mod n.
But gcd(m, n) = 1 implies m mod n 6= 0. Therefore, (i − j) ≡ 0 mod n, or
i ≡ j mod n. Also 0 6 i, j < n forces i = j. Thus no two remainders are same.
But we have n terms in the sequence a, a + m, a + 2m, . . . , a + (n − 1)m and
also for any y, y mod n can have n values, i.e. 0 ≤ y mod n ≤ n − 1. Therefore
invoking pigeonhole principle, we have b mod n must be a remainder for some
i, that is a + im ≡ b mod n. Now let x = a + im, therefore x ≡ a mod m. Also
then x = a + im ≡ b mod n. Thus we have shown there must exist some x,
satisfying x ≡ a mod m and x ≡ b mod n.

14. Using the pigeonhole principle, prove that decimal expansion of a rational
number must, after some time, become repeating.
Solution: Suppose pq be some rational number. We have p = a0 q + r where
0 ≤ r < q. So dividing by q, we have
p r r
= a0 + with 0 ≤ < 1 (1)
q q q
r b1 r1
Again 10r = b1 q + r1 with 0 ≤ r1 < q. Dividing by 10q, we have q = 10 + 10q
r1 1
with 0 ≤ 10q < 10 . Thus we have

p b1 r1 r1 1
= a0 + + with 0 ≤ < (2)
q 10 10q 10q 10
Continuing in the similar fashion, we have
p b1 b2 bn rn rn 1
= a0 + + + · · · + n + n with 0 ≤ n < n (3)
q 10 102 10 10 q 10 q 10

Note that 10rn−1 = bn q + rn with 0 ≤ rn < q. Also (3) implies that decimal
expression of pq is a0 .b1 b2 · · · . So we have 0 ≤ ri < q ∀ i. Now consider the
set {r1 , r2 , · · · , rq+1 }. This set has q + 1 elements with values between −1 and
q. Applying pigeonhole principle, we have rq+1 = ri for some i ≤ q. Thus
the sequence ri must have repetition. Let rm = rn for some m < n. But
10rn = bn+1 q + rn+1 and 10rm = bm+1 q + rm+1 . Unique decomposition of
integers by the Euclidean algorithm implies bn+1 = bm+1 and rn+1 = rm+1 .
Again rn+1 = rm+1 will imply bn+2 = bm+2 and rn+2 = rm+2 . Continuing the
same we get bm+i = bn+i for all i ≥ 0. Thus the decimal expression of pq is
repeating.
Problems (Page 135)
1. If U is an ideal of R and 1 ∈ U , prove that U = R.
Solution: Since we have ur ∈ U ∀ u ∈ U & r ∈ R, so if 1 ∈ U , we have
1r ∈ U ∀ r ∈ R, or r ∈ U ∀ r ∈ R. Therefore R ⊂ U . But by definition
U ⊂ R. Hence U = R.

2. If F is a field, prove its only ideals are (0) and F itself.

Solution: Suppose U be some ideal of F . Now either U = {0} or U 6= {0}.
Clearly U = {0} is an ideal of F . But when U 6= {0}, then there exists some
a ∈ U such that a 6= 0. But F being a field and a 6= 0, therefore there exists a0 ,
inverse of a in F . Now a ∈ U and a0 ∈ F , therefore a.a0 ∈ U , or 1 ∈ U . Again
1 ∈ U and any r ∈ F , therefore 1.r ∈ U , or r ∈ U . Thus F ⊂ U . But U ⊂ F .
So U = F . Thus the only possible ideals of F are {0} or F .

3. Prove that any homomorphism of a field is either an isomorphism or takes

each element into 0.
Solution: Let F be some field and R be some ring. Let φ : F −→ R be some
homomorphism. Let Kφ be kernel of homomorphism φ. We know Kφ is an ideal
of F . But the only ideals of F are {0} or F itself. When Kφ = {0}, we claim φ
is one-to-one mapping. Suppose φ(x) = φ(y) for some x, y ∈ F , then we have
for 0 as an additive identity of R

φ(x) = φ(y) ⇒ φ(x) − φ(y) = 0

⇒ φ(x − y) = 0
⇒ x − y ∈ Kφ
⇒x−y =0
⇒x=y

So when Kφ = {0}, φ is an one-to-one homomorphism (or isomorphism). But

when Kφ = F , then φ(x) = 0 ∀ x ∈ F , or φ takes every element of F into
0. Hence any homomorphism of a field is either an isomorphism or takes each
element into 0.

4. If R is a commutative ring and a ∈ R,

(a) Show that aR = {ar | r ∈ R} is a two-sided ideal of R.
(b) Show by an example that this may be false if R is not commutative.
Solution:
(a) First we will show aR is subgroup of R. Suppose x, y ∈ aR, therefore x = ar1
and y = ar2 for some r1 , r2 ∈ R. But then x − y = ar1 − ar2 = a(r1 − r2 ) = ar3
for some r3 ∈ R. So x − y ∈ aR. Thus aR is subgroup of R under addition.
Next if some x ∈ aR and r ∈ R, then we have x = ar4 for some r4 ∈ R. Also
rx = xr = ar4 r = a(r4 r) = ar5 for some r5 ∈ R. So for all x ∈ aR and r ∈ R,
we have rx, xr ∈ aR. Thus aR is an ideal(or two-sided ideal) of R.
  
a b
(b) Consider R = | a, b, c, d ∈ Z . We left it to the reader to check R
c d  
1 1
is a non-commutative ring. Let a = . Again we can easily check aR =
   0 0  
a b 1 1
| a, b ∈ Z . Clearly, aR is not a two-sided ideal as for ∈ aR
0 0       0 0
1 1 1 1 1 1 1 1
and ∈ R, we have = ∈/ aR. Thus in a non-
1 1 1 1 0 0 1 1
commutative ring R, aR need not to be an ideal.

5. If U, V are ideals of R, let U + V = {u + v | u ∈ U, v ∈ V }. Prove that U + V

is also an ideal.
Solution: Suppose some x, y ∈ U + V , therefore x = u1 + v1 and y = u2 + v2
for some u1 , u2 ∈ U and v1 , v2 ∈ V . But then x − y = (u1 + v1 ) − (u2 + v2 ) =
(u1 − u2 ) + (v1 − v2 ) = u3 + v3 for some u3 ∈ U and v3 ∈ V as U, V are the
ideals of R. So x − y ∈ U + V . Thus U + V is a subgroup of R under addition.
Next suppose some x ∈ U + V and r ∈ R, then we have x = u4 + v4 for some
u4 ∈ U and v4 ∈ V . Now we have xr = (u4 + v4 )r = u4 r + v4 r = u5 + v5 for
some u5 ∈ U and v5 ∈ V as U, V are the ideals of R. So xr ∈ U + V . Similarly
rx ∈ U + V . Thus we have xr, rx ∈ U + V ∀ x ∈ U + V & r ∈ R. So U + V is
an ideal of R.
Remark : We left it to the reader to check U + V is the smallest ideal of R
containing U and V . In other words, hU ∪ V i = U + V .

6. If U, V are ideals of R let U V be the set of all the elements that can be
written as finite sums of elements of the form uv where u ∈ U and v ∈ V . Prove
that U V is an ideal of R.
Solution: We first introduce a change in notation. We assume

U V = {uv | u ∈ U & v ∈ V }
P
Let I = { i∈Λ ui vi | ui ∈ U & vi ∈ V and Λ being some finite index set}.
So wePneed to show I isPan ideal of R. Suppose some x, y ∈ I, therefore
x = i∈Λ ui vi and y = i∈Γ ui vi for ui ∈ U and vi ∈ V for allPi ∈ Λ ∪ Γ,
where Λ, Γ arePsome finite Pindex sets. But Pthen we have x − y = i∈Λ ui vi −
0 0
P
i∈Γ ui vi = i∈Λ ui vi + i∈Γ (−ui )vi = i∈Λ∪Γ ui vi , for some ui ∈ U . So
x − y ∈ I showing I is a P subgroup of R under under addition. Also if some
x ∈ I and r ∈ R, thenPx = i∈∆ ui viPfor all i ∈ ∆, where ∆ is some finite index
set. We have xr = ( i∈∆ ui vi )r = i∈∆ ui vi r = i∈∆ ui vi0 for some vi0 ∈ V .
P
So xr ∈ I for all x ∈ I and r ∈ R. Similarly, rx ∈ I for all x ∈ I and r ∈ R.
Thus I is an ideal of R.
Remark : We left it to the reader to check I is the smallest ideal of R containing
U V . In other words, I = hU V i

7. In Problem 6 prove that U V ⊂ U ∩ V .

Solution: In terms of the notations, we developed in previous problem, we need
to show
hU V i ⊂ U ∩ V
P
Suppose some x ∈ hU V i, therefore x = i∈Γ ui vi for ui ∈ U and vi ∈ V for all
i ∈ Γ, where Γ isP
some finite index set. ButPui vi ∈ U ∀ i ∈ Γ as U is an ideal
of R. Therefore i∈∆ ui vi ∈ U . Similarly, i∈∆ ui vi ∈ V as V too is an ideal
of R. Thus x ∈ U and x ∈ V . Therefore x ∈ U ∩ V . So hU V i ⊂ U ∩ V

8. If R is the ring of integers, let U be the ideal consisting of all multiples of 17.
Prove that if V is an ideal of R and R ⊃ V ⊃ U then either V = R or V = U .
Generalize!
Solution: We have U = 17R and V ideal of R with U ⊂ V ⊂ R. Now either
V = U or U ( V . If U ( V , then there is some x ∈ V such that x ∈ / U . But
x∈/ U implies x 6= 17k for some k ∈ R. But that means 176 | x. Also 17 being a
prime, therefore gcd(17, x) = 1. Therefore 17i + xj = 1 for some i, j ∈ R. But
17i ∈ U ⊂ V and xj ∈ V , therefore 17i + xj ∈ V , or 1 ∈ V . But 1 ∈ V implies
V = R. Hence either V = U or V = R.

We can generalize our result that if p is an irreducible element in R then when-

ever for some ideal V we have pR ⊂ V ⊂ R, implies either V = pR or V = R.

9. If U is an ideal of R, let r(U ) = {x ∈ R | xu = 0 for all u ∈ U } . Prove that

r(U ) is an ideal of R.
Solution: Let some x, y ∈ r(U ), therefore xu = 0 ∀ u ∈ U and yu = 0 ∀ u ∈
U . But then (x − y)u = xu − yu = 0 − 0 = 0 ∀ u ∈ U , therefore im-
plying x − y ∈ r(U ). Thus r(U ) is a subgroup of R under addition. Next
suppose some x ∈ r(U ) and r ∈ R. Therefore x.u = 0 ∀ u ∈ U . We have
(xr)u = x(ru) = x(u1 ) for some u1 ∈ U . Therefore (xr)u = xu1 = 0 ∀ u ∈ U .
So xr ∈ r(U ). Similarly, we can see rx ∈ r(U ). So r(U ) is an ideal of R.

10. If U is an ideal of R let [R : U ] = {x ∈ R | rx ∈ U for every r ∈ R}. Prove

that [R : U ] is an ideal of R and that it contains U .
Solution: Suppose some x, y ∈ [R : U ], therefore rx ∈ U ∀ r ∈ R and
ry ∈ U ∀ r ∈ R. But since U being an ideal, we have rx − ry = r(x − y) ∈
U ∀ r ∈ R, showing x − y ∈ [R : U ]. Thus [R : U ] is a subgroup of R under
addition. Next suppose x ∈ [R : U ] and r1 ∈ R. Therefore rx ∈ U ∀ r ∈ R.
Also r(xr1 ) = (rx)r1 = u1 r1 for some u1 ∈ U . But U being an ideal, there-
fore u1 r1 ∈ U . So r(xr1 ) ∈ U ∀ r ∈ R, implying xr1 ∈ [R : U ] ∀ r1 ∈ R.
Again, r(r1 x) = (rr1 )x = r2 x for some r2 ∈ R. So r(r1 x) = r2 x ∈ U , implying
r1 x ∈ [R : U ]. So r1 x ∈ [R : U ] ∀ r1 ∈ R. Hence [R : U ] is an ideal of R.

Also if x ∈ U , therefore rx ∈ U ∀ r ∈ R as U is an ideal of R. But

rx ∈ U ∀ r ∈ R implies x ∈ [R : U ]. Thus U ⊂ [R : U ].

11. Let R be a ring with unit element. Using its elements we define a ring R̃ by
defining a ⊕ b = a + b + 1, and a · b = ab + a + b, where a, b ∈ R and where the
addition and multiplication on the right-hand side of these relations are those
of R.
(a) Prove that R̃ is a ring under the operations ⊕ and ·.
(b) What act as the zero-element of R̃?
(c) What acts as the unit-element of R̃?
(d) Prove that R is isomorphic to R̃.
Solution:
(a) First note that the both binary operations ⊕ and · are well-defined.

Closure under addition: Since a + b + 1 ∈ R = R̃, therefore a ⊕ b ∈ R̃ for all

a, b ∈ R̃. So R̃ is closed under addition.

Associativity under addition: We have

a ⊕ (b ⊕ c) = a ⊕ (b + c + 1)
= a + (b + c + 1) + 1
= (a + b + 1) + c + 1
= (a ⊕ b) + c + 1
= (a ⊕ b) ⊕ c
Hence associativity under addition holds good.

Existence of additive identity: Suppose e be the additive identity, if it exists.

But then a ⊕ e = a ∀ a ∈ R̃. So a + e + 1 = a ⇒ e = −1 ∈ R̃. So the additive
identity exists and is equal to −1.

Existence of additive inverses: Suppose some a ∈ R̃. If its inverse exists, let it
be a0 . So We have a ⊕ a0 = −1 ⇒ a + a0 + 1 = −1 ⇒ a0 = −2 − a ∈ R̃. So the
inverse element exists for all elements.

Closure under multiplication: We have a · b = ab + a + b ∈ R = R̃. So R̃ is

closed under multiplication.

Associativity under multiplication: We have

a · (b · c) = a · (bc + b + c)
 = a(bc + b + c) + a + (bc + b + c)
= abc + ab + ac + a + bc + b + c
= (ab + a + b)c + (ab + a + b) + c
= (a · b)c + (a · b) + c
= (a · b) · c

Hence R̃ is a ring with ⊕ and · as addition and multiplication respectively.

(b) Already found in part(a), −1 acts as zero-element of R̃.

(c) If exists, let the unity element be u. So we have a · u = a ∀ a ∈ R̃ ⇒

au + a + u = a ⇒ (a + 1)u = 0 ⇒ u = 0 ∈ R̃. Therefore the unity element exists
and is equal to 0.

(d) Define mapping φ : R −→ R̃ such that φ(x) = x − 1. Clearly the mapping

is well-defined. We have
φ(x + y) = (x + y) − 1
= (x − 1) + (y − 1) + 1
= (x − 1) ⊕ (y − 1)
= φ(x) ⊕ φ(y)
Also
φ(xy) = xy − 1
= (x − 1)(y − 1) + (x − 1) + (y − 1)
= (x − 1) · (y − 1)
= φ(x) · φ(y)
So the mapping φ is a ring homomorphism. Also φ(x) = φ(y) ⇒ x − 1 =
y − 1 ⇒ x = y. So φ is one-to-one. Also if some y ∈ R̃, then y + 1 ∈ R is
its inverse-image. So inverse-image of every element exists. So mapping is onto
too. Thus φ is an isomorphism from R onto R̃. And hence R ≈ R̃.

∗
12. In Example 3.1.6 we discussed the ring of rational 2 × 2 matrices. Prove
that this ring has no ideals other than (0) and the ring itself.
Solution: We denote the ring discussed in Example
 3.1.6 as 
M2 (R). 
Suppose U
0 0 0 0
be some ideal of M2 (R). So either U = or U 6= . When
0 0 0 0
     
0 0 a b 0 0
U 6= , we have some A = ∈ U with A 6= .
0 0 c d 0 0
We took a little digression into defining
 another notation
 to represent
 any
 ele-
1 0 0 1 0 0
ment of M2 (R). We define E11 = , E12 = , E21 = , E22 =
0 0 0 0 1 0
 
0 0
. So in this notation, we have A = aE11 + bE12 + cE21 + dE22 . Next we
0 1
claim 
 Eil  if j = k
Eij Ekl = 0 0
if j 6= k
0 0


We left it to the reader to check this result.

 
0 0
Coming back to the problem, we have A 6= . Therefore at least one of
0 0
the a, b, c, d is non-zero. Suppose a 6= 0. So a−1 exists in R. We have

A = aE11 + bE12 + cE21 + dE22

⇒ AE11 = (aE11 + bE12 + cE21 + dE22 )E11
⇒ AE11 = aE11 + cE21
⇒ E11 AE11 = E11 (aE11 + cE21 )
⇒ E11 AE11 = aE11
⇒ (a−1 E11 )AE11 = E11

But (a−1 E11 )AE11 ∈ hAi, therefore

E11 ∈ hAi (1)

⇒ E11 E12 ∈ hAi
⇒ E12 ∈ hAi
⇒ E21 E12 ∈ hAi
⇒ E22 ∈ hAi (2)
⇒ E11 + E22 ∈ hAi Using (1) and (2)
⇒ I2 ∈ hAi
 
1 0
where I2 = is the multiplicative identity of M2 (R). But hAi ⊂ U as
0 1
hAi is the smallest ideal containing A and A ∈ U . Therefore I2 ∈ hAi ⊂ U .
But I2 ∈ U implies U = M2 (R). Also if instead of a some other element, say b
or c or d is non-zero, we can analogously show  I2 ∈ hAi
 ⊂ U , thereby implying
0 0
U = M2 (R). Thus we concluded, either U = or U = M2 (R). Hence
  0 0
0 0
M2 (R) has no ideals other than and M2 (R) itself.
0 0

∗
13. In Example 3.1.8 we discussed the real quaternions. Using this as a model
we define the quaternions over the integers mod p, p an odd prime number,
in exactly the same way; however, now considering all symbols of the form
α0 + α1 i + α2 j + α3 k, where α0 , α1 , α2 , α3 are integers mod p.
(a) Prove that this is a ring with p4 elements whose only ideals are (0) and the
ring itself.
∗∗
(b) Prove that this ring is not a division ring.
Solution:
(a) We denote the quaternions over integers mod p by Qp . It is routine to check
Qp is a ring with 0(≡ 0+0i+0j +0k) as additive identity and 1(≡ 1+0i+0j +0k)
as multiplicative identity. Also o(Qp ) is equal to the number of ways of choosing
four symbols from p symbols with repetition being allowed. Thus o(Qp ) = p4 .
Next we aim to prove {0} and Qp are the only ideals of Qp . Suppose U be
some ideal of Qp . Either U = {0} or U 6= {0}. Suppose U 6= {0}, therefore
some u = a + bi + cj + dk ∈ U with u 6= 0. Since u 6= 0, therefore at least one
of a, b, c, d is non-zero. Suppose a 6= 0, therefore a−1 exists as p being prime
implies Zp is a field. So we have

a + bi + cj + dk ∈ U (1)
⇒ i(a + bi + cj + dk)i ∈ U
⇒ −a − bi + cj + dk ∈ U (2)

Subtracting (2) from (1), we have

2(a + bi) ∈ U
⇒ a + bi ∈ U (3)
⇒ j(a + bi)j ∈ U
⇒ −a + bi ∈ U (4)

Again subtracting (4) from (3), we have

2a ∈ U
⇒a∈U
⇒ aa−1 ∈ U
⇒1∈U

But 1 ∈ U implies U = Qp . In a similar way we can see if b 6= 0 or c 6= 0 or

d 6= 0, we have 1 ∈ U , thereby implying U = Qp . So we conclude if U be some
ideal of Qp , then either U = {0} or U = Qp .

(b) Since p 6= 2, therefore Qp is a non-commutative finite ring. But Wedder-

burn Theorem† asserts that a finite division ring must be commutative. So
Qp must not be a division ring. We can also prove the result using Lagrange
Theorem that any positive integer can be expressed as sum of square of four
integers. So we have p = a2 + b2 + c2 + d2 for some integers a, b, c, d. If
a = b = c = d = 0, then p = 0, which is not the case. So all a, b, c, d can-
not be equal to zero simultaneously. Also if a = b = c = d = 0 mod p, with
† see Section 7.2 of the book
a = b = c = d 6= 0, then p2 | (a2 + b2 + c2 + d2 ), or p2 | p which is not true. So
at least there is some element x out of a, b, c, d such that x 6= 0 mod p. So if
u = a + bi + cj + dk and v = a − bi − cj − dk, then u 6= 0 mod p and v 6= 0 mod
p. But uv = a2 + b2 + c2 + d2 = p = 0 mod p. So we have uv = 0 with neither u
nor v a zero element in Qp . So Qp is not an integral domain, consequently not
a division ring.

If R is any ring a subset L of R is called a left-ideal of R if

1. L is a subgroup under addition.
2. r ∈ R, a ∈ L implies ra ∈ L.
(One can similarly define right-ideal.) An ideal is thus simultaneously a left-
and right-ideal of R.

14. For a ∈ R let Ra = {xa | x ∈ R}. Prove that Ra is a left-ideal of R.

Solution: Suppose x, y ∈ Ra, therefore x = r1 a and y = r2 a for some
r1 , r2 ∈ R. But then x − y = r1 a − r2 a = (r1 − r1 )a = r3 a for some r3 ∈ R. Thus
x−y ∈ Ra. So Ra is a subgroup of R under addition. Next suppose some x ∈ Ra
and r ∈ R. So x = r4 a for some r4 ∈ R. We have rx = r(r4 a) = (rr4 )a = r5 a
for some r5 ∈ R. Thus rx ∈ Ra ∀ x ∈ Ra and r ∈ R. So Ra is a left-ideal of
R.

15. Prove that the intersection of the two left-ideals of R is a left-ideal of R.

Solution: Suppose U1 , U2 be two left-ideals of R. Define U = U1 ∩ U2 . We need
to show U is also a left-ideal of R. Suppose some x, y ∈ U , therefore x ∈ U1 and
x ∈ U2 , y ∈ U1 and y ∈ U2 . Since y ∈ U1 so −y ∈ U1 too as U1 is a left-ideal of
R. So x ∈ U1 and −y ∈ U1 implies x − y ∈ U1 . Similarly x ∈ U2 and −y ∈ U2 ,
implying x − y ∈ U2 . Thus x − y ∈ U1 and x − y ∈ U2 . So x − y ∈ U1 ∩ U2 = U .
Thus U forms a subgroup under addition. Next for x ∈ U and r ∈ R, we have
rx ∈ U1 as U1 is a left-ideal of R. Also rx ∈ U2 as U2 is a left-ideal of R. Thus
rx ∈ U1 and rx ∈ U2 , or rx ∈ U1 ∩ U2 . Thus rx ∈ U ∀ x ∈ U and r ∈ R. So
U is a left-ideal of R.

16. What can you say about the intersection of a left-ideal and right-ideal of
R?
Solution: Intersection of a left-ideal and right-ideal of R need not to a left-
ideal or a 
right-ideal.
 We substantiate  our statement with an example. Consider
a b
M2 (Z) = | a, b, c, d ∈ Z . Clearly M2 (Z) is a ring. We define
c d

  
a 0
U1 = | a, b ∈ Z
b 0
and   
a b
U2 = | a, b ∈ Z
0 0
We left it to the reader to check U1 is a
left-ideal
 of M2 (Z)
 and U2 is a right-
a 0
ideal of M2 (Z). Whereas U1 ∩ U2 = | a ∈ Z . Clearly U1 ∩ U2
0 0
   
1 0 1 1
is not a left ideal as for ∈ U1 ∩ U2 and ∈ M2 (Z), we have
    0 0 1 1
1 1 1 0 1 0
= ∈
/ U1 ∩ U2 . Similarly U1 ∩ U2 is not a right-ideal
1 1 0  0 1 0     
1 0 1 1 1 0 1 1
as for ∈ U1 ∩ U2 and ∈ M2 (Z), we have =
 0 0 1 1 0 0 1 1
1 1
∈
/ U1 ∩ U2 . So U1 ∩ U2 is neither a left-ideal nor a right-ideal.
0 0

17. If R is a ring and a ∈ R let r(a) = {x ∈ R | ax = 0}. Prove that r(a) is a

right-ideal of R.
Solution: Suppose x, y ∈ r(a), therefore ax = ay = 0. But then a(x − y) =
ax − ay = 0 − 0 = 0. So x − y ∈ r(a). Thus r(a) is a subgroup of R under
addition. Next if x ∈ r(a) and r ∈ R, we have a(xr) = (ax)r = 0r = 0. So
xr ∈ r(a) ∀ x ∈ r(a) & r ∈ R. Hence r(a) is a right-ideal of R.

18. If R is a ring and L is a left-ideal of R let λ(L) = {x ∈ R | xa = 0 ∀ a ∈ L}.

Prove that λ(L) is the two-sided ideal of R.
Solution: Suppose x, y ∈ λ(L), therefore xa = 0 ∀a ∈ L and ya = 0 ∀a ∈ L.
But then (x − y)a = 0 ∀ a ∈ L. Thus x − y ∈ λ(L) showing λ(L) a sub-
group of R under addition. Next suppose x ∈ λ(L) and r ∈ R. We have for
a ∈ L, (xr)a = x(ra) = x(a1 ) for some a1 ∈ L as L is the left-ideal of R. So
(xr)a = x(a1 ) = 0 as x ∈ λ(L). Thus xr ∈ λ(L) ∀ x ∈ λ(L) & r ∈ R. Again
for all a ∈ L, (rx)a = r(xa) = r0 = 0. Thus rx ∈ λ(L) ∀x ∈ λ(L) & r ∈ R.
Hence λ(L) is an ideal of R.

19. Let R be a ring in which x3 = x for every x ∈ R. Prove that R is a

commutative ring.
Solution: First suppose x2 = 0 for any x ∈ R. But x2 = 0 ⇒ x(x2 ) = x0 ⇒
x3 = 0 ⇒ x = 0 as x3 = x. Thus

x2 = 0 ⇒ x = 0 (1)

Next, we claim x2 commute with all elements of R. We have

(x2 y − x2 yx2 )2 = (x2 y − x2 yx2 )(x2 y − x2 yx2 )

= x2 yx2 y − x2 yx2 yx2 − x2 yx2 x2 y + x2 yx2 x2 yx2
 = x2 yx2 y − x2 yx2 yx2 − x2 yx4 y + x2 yx4 yx2

But x4 = x3 x = xx = x2 , so

(x2 y − x2 yx2 )2 = x2 yx2 y − x2 yx2 yx2 − x2 yx4 y + x2 yx4 yx2

= x2 yx2 y − x2 yx2 yx2 − x2 yx2 y + x2 yx2 yx2
=0

But then (1) implies x2 y − x2 yx2 = 0 too, or

x2 y = x2 yx2 (2)

Again we can see (yx2 − x2 yx2 )2 = 0. So yx2 − x2 yx2 = 0, or

yx2 = x2 yx2 (3)

From (2) and (3) we conclude x2 y = yx2 ∀x, y ∈ R. Finally, we have for all
x, y ∈ R

xy = xy 3 = (xy 2 )y = y 2 xy
= y 2 x3 y = y 2 x(x2 y) = y 2 xyx2
= yyxyxx = y(yx)(yx)x = y(yx)2 x
= yx(yx)2 = (yx)3 = yx

So xy = yx ∀ x, y ∈ R, showing R to be a commutative ring.

20. If R is a ring with unit element 1 and φ is a homomorphism of R onto R0

prove that φ(1) is the unit element of R0 .
Solution: Let some ȳ ∈ R0 . But since φ is an onto mapping, so there exist
x ∈ R such that φ(x) = ȳ. We have

ȳφ(1) = φ(x)φ(1) = φ(x1) = φ(x) = ȳ

Similarly,
φ(1)ȳ = ȳ
Hence φ(1) is the identity element of R0 .

21. If R is a ring with unit element 1 and φ is a homomorphism of R into an

integral domain R0 such that I(φ) 6= R, prove that φ(1) is the unit element of
R0 .
Solution: Let 0̄ represent the additive identity of R0 . First, we claim φ(1) 6= 0̄.
Suppose φ(1) = 0̄, then we have for all x ∈ R,

φ(x) = φ(x1) = φ(x)φ(1) = φ(x)0̄ = 0̄

But that means I(φ) = R which is not the case. Hence φ(1) 6= 0̄. Also we have

φ(1) = φ(1 · 1) = φ(1)φ(1) (1)

Finally, we claim φ(1) is the multiplicative identity. We establish our claim by

contradiction. Suppose there exist some ȳ ∈ R0 such that ȳφ(1) 6= ȳ. So ȳφ(1) −
ȳ 6= 0̄. Also φ(1) 6= 0̄ and R0 being an integral domain, so (ȳφ(1) − ȳ)φ(1) 6= 0̄.
But

(ȳφ(1) − ȳ)φ(1) 6= 0̄
⇒ ȳφ(1)φ(1) − ȳφ(1) 6= 0̄
⇒ ȳφ(1) − ȳφ(1) 6= 0̄ Using (1)
⇒ 0̄ 6= 0̄, which is not true.

So there exists no such ȳ ∈ R0 such that ȳφ(1) 6= ȳ. Thus ȳφ(1) = ȳ ∀ ȳ ∈ R0 .

Similarly φ(1)ȳ = ȳ ∀ ȳ ∈ R0 Hence φ(1) is the unity element of R0 .
Problems (Page 139)
1. Let R be a ring with unit element, R not necessarily commutative, such that
the only right-ideals of R are (0) and R. Prove that R is a division ring.
Solution: Clearly R 6= {0} as 1 ∈ R. Therefore we can assume some a ∈ R
with a 6= 0. Now consider aR = {ar | r ∈ R}. We can easily prove that aR is a
right-ideal of R. Also it is given that only {0} and R are the only right-ideals of
R. So either aR = {0} or aR = R. Clearly, aR = {0} is not possible as 1 ∈ R,
so a1 = a ∈ aR. So we have only possibility aR = R. Now 1 ∈ R, so for some
a0 ∈ R we have aa0 = 1. But that means, a0 the inverse element of a exists in R.
Since a is some arbitrarily chosen element with the only stipulation that a 6= 0,
so all non-zero elements have inverse in R. Thus R is a division ring.

2. Let R be a ring such that the only right ideals of R are (0) and R. Prove that
either R is a division ring or that R is a ring with a prime number of elements
in which ab = 0 for every a, b ∈ R.
Solution: We define U = {x ∈ R | xr = 0 ∀ r ∈ R} and we claim
U is a right-ideal of R. Clearly 0 ∈ U as 0r = r ∀ r ∈ R. Suppose
u1 , u2 ∈ U , therefore u1 r = 0 ∀ r ∈ R and u2 r = 0 ∀ r ∈ R. But
(u1 − u2 )r = u1 r − u2 r = 0 − 0 = 0 ∀ r ∈ R, therefore u1 − u2 ∈ U . So
U forms a subgroup of R under addition. Also for u ∈ U and r ∈ R, we have
ur = 0 ∈ U . So ur ∈ U ∀ u ∈ U & r ∈ R. Thus U is a right-ideal of R. But
the only right-ideals of R are {0} and R, therefore, either U = {0} or U = R.

Case 1 : When U = R, it means ur = 0 ∀ u ∈ U and r ∈ R i.e. u.r =

0 ∀ u, r ∈ R. Also multiplicative identity, 1 does not exist as if it had ex-
isted would mean 1 ∈ U ⇒ 1R = {0} ⇒ R = {0} ⇒ 1 ∈ / R as 1 6= 0. Also
any subgroup of R under addition is a right-deal as 0 belongs to all subgroup
of R under addition. Therefore {0} and R are the only subgroup of R under
addition. But that mean either R = {0} or o(R) is a prime. Thus in this case
either R = {0} or a ring with prime order, with no multiplicative identity and
satisfying r1 r2 = 0 ∀ r1 , r2 ∈ R. Note when R = {0}, then it is trivially a
division ring.

Case 2 : When U = {0}, it means xr = 0 ∀ r ∈ R only for x = 0. In other

words, for a 6= 0 we have ar 6= 0 at least for some r ∈ R. Now either R = {0}
or R 6= {0}. Suppose R 6= {0}, there exist some a ∈ R with a 6= 0. But
then aR 6= {0}. Also aR is a right-ideal; and {0} and R are the only possible
right-ideals, therefore, aR = R. We claim R to be a division ring. To estab-
lish our claim, we need to show existence of multiplicative right-identity 1 and
right-inverse of any any non-zero element, say a. Suppose some x, y ∈ R such
that xy = 0 with x 6= 0 and y 6= 0. We have xR = R and yR = R as x 6= 0 and
y 6= 0. But then (xy)R = x(yR) = x(R) = R. So xy = 0 ⇒ 0R = R or {0} = R,
which is not the case. Therefore in R, x 6= 0 and y 6= 0 ⇒ xy 6= 0. Reading the
contrapositive of the statement, we have xy = 0 ⇒ x = 0 or y = 0, or R has no
zero-divisors. Now aR = R implies there exist some element u0 ∈ R such that
au0 = a. Clearly u0 6= 0 otherwise that would mean a = 0. Also (au0 )u0 = au0 ,
or a(u0 u0 − u0 ) = 0. But R has no zero-divisors and a 6= 0, so u0 u0 − u0 = 0.
Therefore u0 u0 = u0 . We claim u0 to the required multiplicative right-identity.
Suppose if not, then there must exist some r ∈ R such that ru0 6= r. But then
(ru0 − r)u0 = ru0 u0 − ru0 = ru0 − ru0 = 0, i.e. (ru0 − r)u0 = 0. Again R has no
zero-divisors, so ru0 − r = 0 as u0 6= 0. Thus ru0 = r which is a contradiction.
Hence ru0 = r ∀ r ∈ R, or u0 is the multiplicative right-identity of R. Again
aR = R implies that there exist some a0 such that aa0 = u0 . So the right-inverse
a0 of an arbitrarily chosen element a 6= 0 exists in R. This establishes R to be
a division ring. So we have either R = {0} or is a division ring. But {0} itself
is a division ring. So R is a division ring.

Combining both Cases, we have either R is a division ring or R is a ring of

prime order with r1 r2 = 0 ∀ r1 , r2 ∈ R. Hence the result.

3. Let J be the ring of integers, p a prime number, and (p) the ideal of J
consisting of all multiples of p. Prove
(a) J/(p) is isomorphic to Jp , the ring of integers mod p.
(b) Using Theorem 3.5.1 and part (a) of this problem, that Jp is a field.
Solution:
(a) We define φ : J/hpi −→ Jp such that φ(hpi + x) = x mod p. We claim
mapping φ is well-defined. Suppose some hpi + x = hpi + x0 . So we have
x = x0 + mp for some integer m. Therefore φ(hpi + x) = x mod p = (x0 +
mp) mod p = x0 mod p = φ(hpi + x0 ). Also φ(hpi + x) ∈ Jp ∀ x ∈ J. Thus φ
is well-defined. We have

φ(hpi + x) = φ(hpi + y) ⇒ x ≡ y mod p

⇒ x − y ≡ 0 mod p
⇒ x − y = mp for some integer m
⇒ x − y ∈ hpi
⇒ hpi + x = hpi + y

So φ(hpi + x) = φ(hpi + y) implies hpi + x = hpi + y. Thus mapping φ is one-

to-one. Also if some y ∈ Jp , then we have φ(hpi + y) = y, i.e. every element of
Jp has inverse-image in Jp /hpi. So mapping φ is onto too. Finally, we establish
φ is a homomorphism. We have

φ((hpi + x) + (hpi + y)) = φ(hpi + (x + y))

= (x + y) mod p
= (x mod p + y mod p) mod p
= φ(hpi + x) + φ(hpi + y)
And

φ((hpi + x)(hpi + y)) = φ(hpi + (xy))

= (xy) mod p
= ((x mod p)(y mod p)) mod p
= φ(hpi + x)φ(hpi + y)

So mapping φ is a homomorphism too. Concluding φ is an onto isomorphism

from J/hpi to Jp . So J/hpi ≈ Jp .

(b) First we will show that hpi is a maximal ideal of J. Suppose, if possible
there exists some ideal U of R such that hpi ( U ( J. Since U 6= hpi, therefore
there exists some x ∈ U such that x 6= hpi. So x 6= pk for some integer k. But
that means p6 | x. Also p being prime, therefore gcd(p, x) = 1. Thus pi + xj = 1
for some i, j ∈ J. But p ∈ hpi ⊂ U , therefore pi ∈ U ; also x ∈ U , therefore
xj ∈ U . So pi + xj ∈ U , or 1 ∈ U . But 1 ∈ U implies U = J. Thus there does
not exist an ideal U such that hpi ( U ( J. So hpi is a maximal ideal of J.
Finally, using Theorem 3.5.1, we have J/hpi is a field. But J/hpi ≈ Jp , so Jp
too is a field.

∗∗
4. Let R be the ring of all real-valued continuous functions on the closed unit
interval. If M is a maximal ideal of R, prove that there exists a real number γ,
0 ≤ γ ≤ 1, such that M = Mγ = {f (x) ∈ R | f (γ) = 0}.
Solution:[Warning: solution is wrong explain why!] Let R denotes the field
of real numbers. Suppose M be some maximal ideal of R. We define Aα =
{f (x) |x=α | f (x) ∈ M } where α ∈ [0, 1]. It is easy to see that Aα is an ideal
of R. But R being a field, so either Aα = {0} or Aα = R. Also we have either
Aα = {0} for some α ∈ [0, 1] or Aα 6= {0} for all α ∈ [0, 1].

Case 1 : Suppose Aα = {0} for some α = γ(say) ∈ [0, 1]. Therefore we have
{f (x) |γ | f (x) ∈ M } = {0}. In other words, for all f (x) ∈ M , we have
f (γ) = 0. We define Mγ = {f (x) ∈ R | f (γ) = 0}. Therefore M ⊂ Mγ . So
we have M ⊂ Mγ ⊂ R. Also it is easy to check that Mγ is an ideal of R. But
Mγ 6= R as there are functions h(x) ∈ R such that h(γ) 6= 0. So M being a
maximal implies Mγ = M .

Case 2 : In this case we have Aγ 6= {0} for all α ∈ [0, 1]. Therefore Aα = R
for all α ∈ [0, 1]. Now since M is maximal ideal in R, so there exists a func-
tion g(x) ∈ R such that g(x) ∈ / M . Also for m(x) ∈ M and r(x) ∈ R, we
have m(x)r(x) ∈ M . Therefore g(x) 6= m(x)r(x) for any m(x) ∈ M and any
r(x) ∈ R. But that also means that there exists some β ∈ [0, 1] such that
g(x) |x=β 6= m(x)r(x) |x=β for any m(x) ∈ M and any r(x) ∈ R. Also Aβ = R,
so m(β) can assume any value in R. Also r(β) can assume any value in R.
Therefore m(β)r(β) can assume any value in R. So g(β) 6= m(β)r(β) is not
possible as g(β) ∈ R. So Aα = R for all α ∈ [0, 1] is not possible.

Thus we concluded if M is some maximal ideal of R, then M must equal to Mγ

for some γ ∈ [0, 1], where Mγ = {f (x) ∈ R | f (γ) = 0}.
Problems (Page 142)
1. Prove that if [a, b] = [a0 , b0 ] and [c, d] = [c0 , d0 ] then [a, b][c, d] = [a0 , b0 ][c0 , d0 ].
Solution: We have

[a, b] = [a0 , b0 ] ⇔ ab0 = a0 b (1)

Similarly,

[c, d] = [c0 , d0 ] ⇔ cd0 = c0 d (2)

We need to show

[a, b][c, d] = [a0 , b0 ][c0 , d0 ]

⇔ [ac, bd] = [a0 c0 , b0 d0 ]
⇔ acb0 d0 = a0 c0 bd (3)

We have

acb0 d0 = (ab0 )(cd0 )

= (a0 b)(c0 d)

Using (1) and (2)

= a0 c0 bd

Hence [a, b][c, d] = [a0 , b0 ][c0 , d0 ].

2. Prove the distributive law in F .

Solution: We have

[a, b]([c, d] + [e, f ]) = [a, b][cf + ed, df ]

= [a(cf + ed), bdf ]
= [acf + aed, bdf ]
= [bacf + baed), b2 df ]
= [(ac)(bf ) + (ae)(bd), (bd)(bf )]
= [ac, bd] + [ae, bf ]
= [a, b][c, d] + [a, b][e, f ]

Similarly the other distributive law hold good.

3. Prove that the mapping φ : D −→ F defined by φ(a) = [a, 1] is an isomor-

phism of D into F .
Solution: We need to show φ is an one-to-one homomorphism. Clearly map-
ping φ is well-defined. Also we have

φ(a + b) = [a + b, 1]
= [a, 1] + [b, 1]
= φ(a) + φ(b)

Also

φ(ab) = [ab, 1]
= [a, 1][b, 1]
= φ(a)φ(b)

So mapping φ is a ring homomorphism. Also we have

φ(a) = φ(b) ⇒ [a, 1] = [b, 1]

⇒ a1 = b1
⇒a=b

Thus mapping φ is one-to-one too. Hence φ is an one-to-one homomorphism.

4. Prove that if K is any field which contains D then K contains a subfield

isomorphic to F . (In this sense F is the smallest field containing D.)
Solution: We are given D is an integral domain; K some field containing D;
and F field of quotients of D. We define φ : F −→ K such that φ([a, b]) = ab−1 .
We claim mapping φ so defined is a well-defined mapping. We have [a, b] ∈
F ⇒ a, b ∈ D with b 6= 0 ⇒ a, b ∈ K with b 6= 0 ⇒ a, b−1 ∈ K ⇒ ab−1 ∈ K.
Also if [a, b] = [a0 , b0 ] with b, b0 6= 0, then we have ab0 = a0 b ⇒ ab0 (b−1 b0−1 ) =
a0 b(b−1 b0−1 ) ⇒ ab−1 = a0 b0−1 ⇒ φ([a, b]) = φ([a0 , b0 ]). Hence the mapping is
well-defined.
Also we have

φ([a, b] + [c, d]) = φ([ad + cb, bd])

= (ad + cb)(bd)−1
= (ad + cb)(b−1 d−1 )
= adb−1 d−1 + cbb−1 d−1
= ab−1 + cd−1
= φ([a, b]) + φ([c, d])

and

φ([a, b][c, d]) = φ([ac, bd])

 = ac(bd)−1
= acb−1 d−1
= (ab−1 )(cd−1 )
= φ([a, b])φ([c, d])

Thus φ is a ring homomorphism. Also we have

φ([a, b]) = φ([c, d]) ⇒ ab−1 = cd−1

⇒ ab−1 (bd) = cd−1 (bd)
⇒ ad = cb
⇒ [a, b] = [c, d]

So mapping φ is one-to-one too. Thus φ is an one-to-one homomorphism. Also

φ(F ) is a subfield of K (Check). Thus F ≈ φ(F ). Hence the result.

∗
5. Let R be a commutative ring with unit element. A non-empty subset S of
R is called a multiplicative system if
1. 0 ∈/S
2. s1 , s2 ∈ S implies that s1 , s2 ∈ S.
Let M be the set of all ordered pairs (r, s) where r ∈ R, s ∈ S. In M define
(r, s) ∼ (r0 , s0 ) if there exists an element s00 ∈ S such that

s00 (rs0 − sr0 ) = 0.

(a) Prove that this defines an equivalence relation on M.

Let the equivalence class of (r, s) be denoted by [r, s], and let RS be the set of
all the equivalence classes. In RS define [r1 , s1 ] + [r2 , s2 ] = [r1 s2 + r2 s1 , s1 s2 ]
and [r1 , s1 ][r2 , s2 ] = [r1 r2 , s1 s2 ].
(b) Prove that the addition and multiplication described above are well-defined
and that RS forms a ring under these operations.
(c) Can R be embedded in RS ?
(d) Prove that the mapping φ : R −→ RS , defined by φ(a) = [as, s] is a homo-
morphism of R into RS and find the kernel of φ.
(e) Prove that this kernel has no element of S in it.
(f) Prove that every element of the form [s1 , s2 ](where s1 , s2 ∈ S) in RS has an
inverse in RS .
Solution:
(a) A relation is an equivalence if it satisfies reflexivity, symmetry and transi-
tivity properties.
Reflexivity: We have s0 (rs − rs) = 0 for any s0 ∈ S, which means (r, s) ∼ (r, s).
Hence the relation is reflexive.
Symmetry: We are given (r, s) ∼ (r0 , s0 ). So s00 (rs0 − sr0 ) = 0 for some s00 ∈ S.
But s00 (rs0 −sr0 ) = 0 ⇒ −s00 (r0 s−s0 r) = 0 ⇒ (r0 , s0 ) ∼ (r, s). Hence the relation
is symmetric too.
Transitivity: We are given (r, s) ∼ (r0 , s0 ) and (r0 , s0 ) ∼ (r00 , s00 ). But (r, s) ∼
(r0 , s0 ) implies s1 (rs0 − r0 s) = 0 for some s1 ∈ S. So we have

s1 rs0 = s1 r0 s (1)

Similarly, (r0 , s0 ) ∼ (r00 , s00 ) implies, for some s2 ∈ S

s2 r0 s00 = s2 r00 s0 (2)

We need to show

(r, s) ∼ (r00 , s00 )

⇔ s3 (rs00 − r00 s) = 0

for some s3 ∈ S. Let s3 = s1 s2 s0 , therefore

(r, s) ∼ (r00 , s00 )

⇐ s1 s2 s0 (rs00 − r00 s) = 0
⇐ (s1 rs0 )s2 s00 − s1 s2 s0 r00 s = 0

Using (1),

⇐ (s1 r0 s)s2 s00 − s1 s2 s0 r00 s = 0

⇐ s1 s(s2 r0 s00 ) − s1 s2 s0 r00 s = 0

Using (2),

⇐ s1 s(s2 r00 s0 ) − s1 s2 s0 r00 s = 0

⇐0=0

Thus the relation is transitive too. Hence the relation ∼ is an equivalence rela-
tion.

(b) We will first show addition + : RS × RS −→ RS is well-defined. Suppose

[r1 , s1 ] = [r10 , s01 ] and [r2 , s2 ] = [r20 , s02 ]. But [r1 , s1 ] = [r10 , s01 ] implies

s3 r1 s01 = s3 r10 s1 , (3)

for some s3 ∈ S. Also [r2 , s2 ] = [r20 , s02 ] implies

s4 r2 s02 = s4 r20 s2 , (4)

for some s4 ∈ S. Now in order to prove addition is well-defined, we need to

show

[r1 , s1 ] + [r2 , s2 ] = [r10 , s01 ] + [r20 , s02 ]

 ⇔ [r1 s2 + r2 s1 , s1 s2 ] = [r10 s02 + r20 s01 , s01 s02 ]
⇔ s5 ((r1 s2 + r2 s1 )s01 s02 − (r10 s02 + r20 s01 )s1 s2 ) = 0

for some s5 ∈ S. Let s5 = s3 s4 , therefore

[r1 , s1 ] + [r2 , s2 ] = [r10 , s01 ] + [r20 , s02 ]

⇐ s3 s4 ((r1 s2 + r2 s1 )s01 s02 − (r10 s02 + r20 s01 )s1 s2 ) = 0

But putting values form (3) and (4), we have left-hand side of the above equa-
tion equals to 0. Hence addition is well-defined.

Next we will show multiplication is well-defined. Again suppose [r1 , s1 ] = [r10 , s01 ]
and [r2 , s2 ] = [r20 , s02 ]. But these equalities imply (3) and (4) respectively. We
need to show

[r1 , s1 ][r2 , s2 ] = [r10 , s01 ][r20 , s02 ]

⇔ [r1 r2 , s1 s2 ] = [r10 r20 , s01 s02 ]
⇔ s5 (r1 r2 s01 s02 − r10 r20 s1 s2 ) = 0

for some s5 ∈ S. Let s5 = s3 s4 , therefore

[r1 , s1 ][r2 , s2 ] = [r10 , s01 ][r20 , s02 ]

⇐ s3 s4 (r1 r2 s01 s02 − r10 r20 s1 s2 ) = 0

Putting values from (3) and (4), we have left-hand side of the above equation
equals to 0. Hence multiplication is well-defined too.

Since S is assumed to be non-empty, so some s0 ∈ S. Now rest is routine to

check RS is a commutative ring with unity. Note that [0, s0 ] is the additive
identity and [s0 , s0 ] is the multiplicative identity.

(c) In general no. (See part (d))

(d) We have φ : R ←− RS such that φ(a) = [as, s]. Easy to check φ is well-
defined. Also

φ(a + b) = [(a + b)s, s]

= [as2 + bs2 , ss]
= [as, s] + [bs + s]
= φ(a) + φ(b)

and

φ(ab) = [(ab)s, s]
 = [(as)(bs), ss]
= [as, s][bs, s]
= φ(a)φ(b)

Hence φ is a ring homomorphism.

Let Kφ denotes the kernel of mapping φ. Now Kφ in general depends upon
the choice of S. We illustrate it with example. Suppose some positive integer
n = pq, where p and q are prime integers. Consider R = Zn with addition
modulo n and multiplication modulo n as its addition and multiplication. Let
Un = {1 ≤ x < n | gcd(x, n) = 1}. Let S = Un and S 0 = Un ∪ {p}. One
can check both S and S 0 are multiplicative system of R. But when we are
working with the multiplicative system S, Kφ turns out to be {0}. Whereas
when we have S 0 as our multiplicative system, with φ(a) = [ap, p], Kφ 6= {0} as
φ(q) = [0, p]. As for a given ring, there might be more than one possible multi-
plicative system, so Kφ , in general depends upon the multiplicative system that
we have chosen.

(e) Let φ : R −→ RS such that φ(a) = [as, s], where s ∈ S. Let some x ∈ Kφ ,
therefore φ(x) = [0, s] ⇒ [xs, s] = [0, s] ⇒ s0 xs2 = 0 for some s0 ∈ S. Now if
x ∈ S, then s0 xs2 6= 0 ∀ s0 ∈ S. Thus if x ∈ Kφ , then x ∈/ S. Hence the result.

(f) We have [s2 , s1 ] as the inverse element of [s1 , s2 ] for s1 , s2 ∈ S as [s2 , s1 ][s1 , s2 ] =
[s, s] where s ∈ S and [s, s] is the multiplicative identity. Hence every element
of form [s1 , s2 ] for s1 , ss ∈ S has inverse in RS .

6. Let D be an integral domain, a, b ∈ D. Suppose that an = bn and am = bm

for two relatively prime positive integers m and n. Prove that a = b.
Solution: First note that D is given only an integral domain, therefore multi-
plicative identity and inverse of an element under multiplication may not exist.
So for x ∈ D, xn is defined only for positive integers n.

When a = 0, we have bn = an = 0n = 0 where n is given a positive integer.

If n = 1, then b1 = 0 = a, hence the result. But if n > 1, then also we claim
bn = 0 implies b = 0. Suppose we have bn = 0 with b 6= 0. But then there must
exist 1 < p ≤ n such that bp = 0 and bp−1 6= 0. So we have bp = bp−1 b = 0,
also D being an integral domain implies b = 0 or bp−1 = 0. In both the cases
we have contradiction. Therefore bn = 0 ⇒ b = 0. So if a = 0, then a = b.

When a 6= 0, we will prove the result by making use of the field of quotients
of D. Let F be the field of quotients of D. Define φ : D −→ F such that
φ(x) = [xd, d], where d 6= 0 ∈ D. It is easy to check that φ so defined is an
one-to-one ring homomorphism. Now in the field F , with a 6= 0 and d 6= 0, we
have [ad, d] 6= [0, d] as D has no zero-divisors. So [ad, d]k is well-defined for all
k ∈ Z. So we have

φ(a) = [ad, d]
= [ad, d]1
= [ad, d]mi+nj as gcd(m, n) = 1
= [(ad)mi+nj , dmi+nj ]
= [ami+nj dmi+nj , dmi+nj ] as D is commutative
= [ami+nj d, d]
= [ami anj d, d]
= [(am )i (an )j d, d]
= [(bm )i (bn )j d, d]
= [bmi+nj d, d]
= [bd, d]
= φ(b)

But mapping φ being one-to-one, so φ(a) = φ(b) implies a = b. Hence the

result.

7. Let R be a ring, possibly noncommutative, in which xy = 0 implies x = 0 or

y = 0. If a, b ∈ R and an = bn and am = bm for two relatively prime positive
integers m and n, prove that a = b.
Solution: First note that R may not have multiplicative identity or the inverse
element of all elements. So for x ∈ R, xl is only defined for positive integers l.
Next we claim that in R, xl = 0 for some positive integer l implies x = 0. Sup-
pose x 6= 0. But xl = 0, therefore there exists some positive integer p > 1 such
that xp = 0 and xp−1 6= 0. But then we have xp = xp−1 x = 0, R being having
no zero-divisors, implying xp−1 = 0 or x = 0, both of which is a contradiction.
Hence xl = 0 ⇒ x = 0.

Next since m and n are relatively prime, so mi + nj = 1 for some i, j ∈ Z. First

suppose i > 0, therefore nj = 1 − mi < 0 as m > 1. Note if m or n is equal to
1 then we have nothing to prove, so we have assumed m, n > 1. Therefore for
i > 0, we have j < 0. Let j = −k, therefore k > 0 and mi − nk = 1. So we have

ami = a1+nk
⇒ (am )i = a(an )k
⇒ (bm )i = a(bn )k
⇒ bmi b = abnk b
⇒ bmi+1 = abnk+1
⇒ bbmi = abmi
 ⇒ (b − a)bmi = 0

So R being having no zero-divisors, we have b − a = 0 or bmi = 0. When

b − a = 0, we have b = a. While bmi = 0 implies b = 0, which in turn mean
an = bn = 0n = 0. So a = 0 too. Thus a = b in both cases. Similarly when
i < 0, we can show j > 0 and can proceed in a similar fashion to prove a = b.
Finally if i = 0, then we have nj = 1 ⇒ n = 1 ⇒ a1 = b1 . So a = b in this case
too. Hence a = b.
Problems (Page 149)
1. In a commutative ring with unit element prove that the relation a is associate
of b is an equivalence relation.
Solution: A relation is an equivalence relation if it satisfies reflexivity, symme-
try, and transitivity properties.

Reflexivity: Since a = 1a, and 1 is also a unit element, so a is associate of

a itself. Thus associate relation is reflexive.

Symmetry: Suppose a is associate of b. So a = ub for some unit element u.

u being unit element, therefore u−1 exists. So we have a = ub ⇒ b = u−1 a,
showing b is associate of a. Thus the associate relation is symmetric too.

Transitivity: Suppose a is associate of b, and b is associate of some c.

So we have a = u1 b and b = u2 c for some units u1 and u2 . Therefore
a = u1 b = u1 u2 c. But u1 u2 is again a unit as (u1 u2 )(u−1 −1
1 u2 ) = 1. So a
is associate of c. Thus the associate relation is transitive too.

And hence the associate relation is an equivalence relation.

2. In a Euclidean ring prove that any two greatest common divisors of a and b
are associates.
Solution: Suppose d1 and d2 are greatest common divisors of a and b. That
means d1 | a and d1 | b. But d2 being greatest common divisor of a and b, there-
fore if some d1 | a and d2 | b, then d2 must divide d1 . So d2 | d1 . Symmetry of
the argument implies d1 | d2 . But then by Lemma 3.7.2, we have d1 and d2 are
associates. Hence any two greatest common divisors are associates.

3. Prove that a necessary and sufficient condition that the element a in the
Euclidean ring be a unit is that d(a) = d(1).
Solution: First, suppose a is a unit element. We will show d(a) = d(1). Since
in a Euclidean ring, d(b) ≤ d(ba) for all non-zero a and b, so assuming b = 1,
we have d(1) ≤ d(a1), or d(1) ≤ d(a). Also a being unit element, so a−1 exists.
Again d(a) ≤ d(ab), so putting b = a−1 , we have d(a) ≤ d(aa−1 ) ⇒ d(a) ≤ d(1).
Hence d(a) = d(1).

Conversely, suppose for some non-zero a, d(a) = d(1). We need to show a is a

unit element. In a Euclidean ring, we have 1 = qa + r for some q and r, with
either r = 0 or d(r) < d(a). When r = 0, we have 1 = qa, which means a is a
unit element. When r 6= 0, we have d(r) < d(a). But d(a) = d(1), so

d(r) < d(1) (1)

Also d(1) ≤ d(1r), i.e.

d(1) ≤ d(r) (2)
But (1) and (2) implies d(r) < d(r) which is absurd, hence r = 0 is the only
possibility. So a is a unit element. Thus in a Euclidean ring, a is a unit element
if and only if d(a) = d(1).

4. Prove that in a Euclidean ring (a, b) can be found as follows:

b = q0 a + r1 , where d(r1 ) < d(a)

a = q1 r1 + r2 , where d(r2 ) < d(r1 )
r1 = q2 r2 + r3 , where d(r3 ) < d(r2 )
.. ..
. .
rn−1 = qn rn
and rn = (a, b).

Solution: We will first show that gcd(a, b) = gcd(a, b − qa) for all q ∈ R,
where R is assumed to be a Euclidean ring. Suppose d1 = gcd(a, b) and
d2 = gcd(a, b − qa). So d1 | a and d2 | b. Therefore d1 | a and d1 | (b − qa).
But gcd(a, b − qa) = d2 , so d1 | d2 . Again d2 | a and d2 | (b − qa). Therefore d2 | a
and d2 | b. But gcd(a, b) = d1 . Therefore d2 | d1 . But d1 | d2 and d2 | d1 implies
d1 and d2 are associates. So gcd(a, b) = gcd(a, b − qa) upto associates.

Now since R is a Euclidean ring, therefore b = q0 a + r1 , where either r1 = 0 or

d(r1 ) < d(a). If r1 = 0, we are done as a is the required gcd. But if r1 6= 0,
then we have gcd(a, b) = gcd(a, b − q0 a) = gcd(a, r1 ) = gcd(r1 , a). Again we
write a = q1 r1 + r2 for some q1 , r2 with either r2 = 0 or d(r2 ) < d(r1 ). Again
if r2 = 0, then clearly gcd(a, b) = gcd(r1 , a) = r1 . But if r2 6= 0, then we have
gcd(a, b) = gcd(r1 , a) = gcd(r2 , r1 ). We can continue like this till we get some
rn+1 = 0. Also when rn+1 = 0, we have gcd(a, b) = gcd(r1 , a) = gcd(r2 , r1 ) =
· · · = gcd(rn , rn−1 ) = rn as rn−1 = qn rn + 0. All left is to show that rn+1
must be equal to 0 for some n ∈ N. But suppose the process keep on going in-
finitely with all ri not equal to 0. But then we get a strictly decreasing sequence
d(a), d(r1 ), d(r2 ), · · · in N as d(a) > d(r1 ) > d(r2 ) > · · · . Also d(ri ) ≥ 0 ∀ i.
So rn+1 must be equal to 0 for some n ∈ N. Hence the result.

5. Prove that if an ideal U of a ring R contains a unit of R, then U = R.

Solution: We will make use of the fact that if some u ∈ U and r ∈ R, then
ur ∈ U . Suppose U contains some unit element u. As u is a unit, therefore u−1
exists in R. Now u ∈ U and u−1 ∈ R, therefore uu−1 ∈ U , or 1 ∈ U . Again
1 ∈ U and let some r ∈ R, so we have 1r ∈ U , or r ∈ U . That is for all r ∈ R,
we have r ∈ U , which means R ⊂ U . But by definition, U ⊂ R. Therefore
U = R. Hence the result.
6. Prove that the units in a commutative ring with a unit element form an
abelian group.
Solution: Let I be the set of all units elements. Ring being commutative
implies I is commutative under multiplication. Suppose u1 and u2 are units.
Therefore u−1 −1
1 , u2 exist. But then (u1 u2 )(u−1 −1
1 u2 ) = 1. Therefore u1 u2 ∈ I,
or the closure property holds good. I being subset of the ring, therefore asso-
ciativity under multiplication holds for all its elements. Also 1, multiplicative
identity being a unit belongs to I too. Finally suppose some u1 ∈ I, therefore
there exists u−1 −1
1 in the ring such that u1 u1 = 1. But the equation also tells us
−1 −1
that u1 is a unit element, or u1 ∈ I. Therefore existence of inverse of each
element is also shown. So I is an abelian group under the multiplication.

7. Given two elements a, b in the Euclidean ring R their least common multiple
c ∈ R is an element in R such that a | c and b | c and such that whenever a | x
and b | x for x ∈ R then c | x. Prove that any two elements in the Euclidean ring
R have a least common multiple in R.
Solution: We assume both a and b as non-zero elements, otherwise a | c or b | c
would not be defined. We define hai as the smallest ideal of R containing a.
One can easily see that whenever some ring R is commutative and has unity
element, then we have hai = aR = {ar | r ∈ R}. So hai = aR. Similarly we
have hbi = bR. Let U = hai ∩ hbi. Clearly U is an ideal of R. Now we make use
of the fact that R is a principal ideal ring. Note that a Euclidean ring is always
a principal ideal ring. So U = cR for some c ∈ R. We claim c is the required
least common multiple of a and b. We have U ⊂ hai. Also c = c1 ∈ cR. So
c ∈ U ⊂ hai. Therefore c = ar1 for some r1 ∈ R, or a | c. Similarly, c ∈ U ⊂ hbi,
implying c = br2 , or b | c. So a | c and b | c. Next suppose a | x and b | x for some
x ∈ R. Therefore x = ar3 and x = br4 for some r3 , r4 ∈ R. But that would
mean x ∈ aR = hai and x ∈ bR = hbi. So x ∈ hai ∩ hbi = U . Therefore x = cr5
for some r5 ∈ R. So c | x. Thus whenever a | x and b | x implies c | x. Thus c is
the least common multiple of a and b. So we concluded that the least common
multiple of two non-zero elements always exists in a Euclidean ring.

8. In Problem 7, if the least common multiple of a and b is denoted by [a, b],

prove that [a, b] = ab/(a, b).
Solution: Let c and d be the least common multiple and the greatest common
divisor respectively of a and b in R. We assume R to be a Euclidean ring.
Therefore hci = hai ∩ hbi and hdi = ha, bi = hai + hbi. Also in a commutative
ring we have hxyi = hhxihyii. So we have

hcdi = hhcihdii
= h(hai ∩ hbi)(hai + hbi)i
= h(hai ∩ hbi)hai + (hai ∩ hbi)hbii
= h((haihai) ∩ (hbihai)) + ((haihbi) ∩ (hbihbi))i
 = h(hai ∩ (haihbi)) + ((haihbi) ∩ hbi)i
= hhaihbi + haihbii
= hhaihbii
= habi

But hcdi = habi implies cd = ab(upto associates). Hence the result.

Problems (Page 152)
1. Find all the units in J[i].
Solution: Using Problem 3 (Page 149 of the book), we have u a unit element
of J[i] if and only if d(u) = d(1). Let u = a + bi, therefore u is a unit element
if and only if d(u) = a2 + b2 = d(1) = 12 + 02 = 1. But the integral solutions of
a2 + b2 = 1 are a = 0, b = ±1 and a = ±1, b = 0. Thus i, −i, 1, −1 are the only
unit elements of J[i]

2. If a + bi is not a unit of J[i] prove that a2 + b2 > 1.

Solution: We have d(a + bi) ∈ W. If d(a + bi) = a2 + b2 = 0, then a + bi = 0.
When d(a + bi) = 1 = d(1), then a + bi is a unit element. So if a + bi is neither
a unit element nor a zero element then d(a + bi) > 1. Hence the result.

3. Find the greatest common divisor in J[i] of

(a) 3 + 4i and 4 − 3i (b) 11 + 7i and 18 − i.
Solution:
(a) Clearly 3 + 4i = i(4 − 3i). So gcd(3 + 4i, 4 − 3i) = 4 − 3i

(b) We resort to Q[i]. We have

18 − i (18 − i)(11 − 7i)
=
11 + 7i (11 + 7i)(11 − 7i)
191 − 137i
=
 170   
21 33
= 1+ − 1− i
170 170
21 + 33i
= (1 − i) + = q + r0 (say) (1)
170
Note that we have reduced r0 = a + ib (say) such that |a|, |b| ≤ 21 . So d(r0 ) ≤
( 21 )2 + ( 12 )2 < 1. Multiplying (1) by 11 + 7i, we have 18 − i = (11 + 7i)q + (11 +
7i)r0 = (11 + 7i)q + r. So d(r) = d((11 + 7i)r0 ) = d(11 + 7i)d(r0 ) < d(11 + 7i)
as d(r0 ) < 1. Thus we are following the steps of Euclidean algorithm for finding
gcd as described in Problem 4 (Page 149 of the book). So we have 18 − i =
(11 + 7i)(1 − i) + r where r can be found by equating both sides of this equation
itself. Thus we have 18 − i = (11 + 7i)(1 − i) + 3i with d(3i) < d(11 + 7i). Also
gcd(18 − i, 11 + 7i) = gcd(11 + 7i, 3i)
Again
11 + 7i (11 + 7i)(−3i)
=
3i (3i)(−3i)
21 − 33i
=
9
 3 + 3i
= (2 − 4i) +
9

So we have 11 + 7i = (3i)(2 − 4i) + (−1 + i) and

gcd(11 + 7i, 3i) = gcd(3i, −1 + i)

Again we have

3i 3i(−1 − i)
=
−1 + i (−1 + i)(−1 − i)
3 − 3i
=
2
1−i
= (1 − i) +
2
So we have 3i = (−1 + i)(1 − i) + i and

gcd(3i, −1 + i) = gcd(−1 + i, i)

But gcd(−1 + i, i) = 1 as i is a unit element of J[i]. Hence

gcd(18 − i, 11 + 7i) = 1

4. Prove that if p is a prime number of the form 4n + 3, then there no x such

that x2 ≡ −1 mod p.
Solution: We need to show there is no x such that x2 + 1 ≡ 0 mod p, when p
is of form 4n + 3. Consider polynomial f (x) = x2 + 1 in Z4 [x]. We have

f (x) |x=0 = 1
f (x) |x=1 = 2
f (x) |x=2 = 1
f (x) |x=3 = 2

Thus f (x) 6= 3 in Z4 [x]. Now consider f (x) as a polynomial in Z[x]. So we

have f (x) 6= 4n + 3 for all x ∈ Z and for any n ∈ Z. So if p is of form 4n + 3,
f (x) = x2 + 1 6≡ 0 mod p for any x ∈ Z. Hence the result.

ALITER: Suppose x2 ≡ −1 mod p has solution, where p = 4n + 3 for some

n ∈ W. We have p−1
2 = 2n + 1. So we have

p−1 p−1
x2 ≡ −1 mod p ⇒ (x2 ) 2 = (−1) 2 mod p
p−1 2n+1
⇒x = (−1) mod p
But we have xp−1 = 1 mod p, for p a prime number (Fermat Theorem), so

x2 ≡ −1 mod p ⇒ 1 = −1 mod p

which is only possible when p = 2, which is not case. Hence x2 ≡ −1 mod p has
no solution in x for p prime of form 4n + 3.

5. Prove that no prime of the form 4n + 3 can be written as a2 + b2 where a

and b are integers.
Solution: As in the previous problem we consider a2 + b2 with a, b ∈ Z4 . By
brute force, we can see a2 +b2 6= 3 mod 4 for all a, b ∈ Z4 [x]. Considering a2 +b2
in Z, we have a2 + b2 6= 4n + 3 for any n ∈ Z. Thus if p is of form 4n + 3, then
it cannot be equal to a2 + b2 for any a, b ∈ Z. Hence the result.

6. Prove that there is an infinite number of primes of the form 4n + 3.

Solution: We will adapt the proof given by Euclid. Suppose primes of form
4n+3 are finite. Let 3 = p1 < p2 < · · · < pk for some k ∈ N, are the all primes of
form 4n+3. Consider a = 4(p1 p2 · · · pk )−1. Clearly, a = 4((p1 p2 · · · pk )−1)+3,
i.e. is a number of form 4n + 3. Also a > pk , so a is a composite number as
p1 , p2 , · · · , pk are the only primes of form 4n + 3. We have pi 6 | a ∀ i as if pi | a
implies pi | 1 which is not the case. Also 26 | a as 26 | 1. Thus if a is a composite
number then a = q1 q2 · · · ql where qi ∀ i are primes of form 4n + 1. But prod-
uct of integers of form 4n + 1 is again an integer of form 4n + 1. Thus it leads
to the conclusion that a is an integer of form 4n + 1, which is not true as a is
an integer of form 4n + 3. Thus primes of form 4n + 3 cannot be finite.

∗
7. Prove there exists an infinite number of primes of the form 4n + 1.
Solution: Suppose there are finite number of primes of form 4n + 1. Let
p1 , p2 , · · · , pk be all the primes of form 4n + 1 for some k ∈ N. We define
a = (2p1 p2 · · · pk )2 + 1. Clearly a is of form 4n + 1. But a > pi for all i, there-
fore a must be composite. Let some prime p divides a. Therefore a = 0 mod
p ⇒ (2p1 p2 · · · pk )2 + 1 = 0 mod p ⇒ (2p1 p2 · · · pk )2 = −1 mod p. This means
that the equation x2 = −1 mod p in x has solution which is x = 2p1 p2 · · · pk .
But then by Problem 4, p must not be of form 4n + 3. So p is of form 4n + 1
or is equal to 2. When p = 2, we have 2 | a ⇒ 2 | (2p1 p2 · · · pk )2 + 1. But
2 | (2p1 p2 · · · pk )2 , so 2 | a implies 2 | 1 which is not true. So p = 2 is not feasible.
When p is of form 4n + 1, then p = pi for some i as these are the only primes of
form 4n + 1. But then again p | a leads to conclusion that p | 1 which is not true.
Thus the assumption that primes of form 4n+1 are finite leads to contradiction.
Hence primes of form 4n + 1 must be infinite in number.

∗
8. Determine all the prime elements in J[i].
Solution: We first claim if z ∈ J[i] is a prime element(unique upto asso-
ciates), then z | p for some prime p ∈ Z. To establish our claim first note
that z z̄ = d(z) ∈ N. But Z being a Unique Factorization Domain, so we
have z z̄ = d(z) = p1 p1 · · · pk for some prime elements pi ∈ Z. Also z | z z̄, so
z | p1 p2 · · · pk . Now z a prime element in J[i] and pi belonging in J[i] too. So z
must divides some pi . Hence our claim. We restate our result again that if z is
prime in J[i], then it must divides some prime element of Z. So to categorize all
prime elements of J[i], we first categorize prime elements in Z and then find all
prime elements in J[i] corresponding to each prime in Z. We categorize prime
elements of Z in three categories; prime elements of form 4n + 1, prime elements
of form 4n + 3, and the prime element 2.

Case 1 p is a prime element of form 4n + 1: We have by Theorem 3.8.2,

p = a2 + b2 = (a + bi)(a − bi). So d(p) = d((a + bi)(a − bi)) = d(a + bi)d(a − bi) =
(a2 + b2 )(a2 + b2 ). So p2 = (a2 + b2 )2 , or p = a2 + b2 = d(a + bi) = d(a − bi).
But d(a + ib) = p, a prime element in Z implies a + ib is a prime element in J[i]
as if a + bi is not a prime element in J[i] mean a + bi = z1 z2 with d(z1 ) 6= 1 and
d(z2 ) 6= 1, which in turn implies p = d(a + bi) = d(z1 z2 ) = d(z1 )d(z2 ), showing p
is not a prime element in Z which is not true. Similarly, a−bi is a prime element
in J[i]. Next, we claim a + bi and a − bi are not associates because if a + bi and
a − bi are associates that would imply a = ±b ⇒ a2 = b2 ⇒ p = 2a2 ⇒ 2 | p
which is not the case. So a + bi and a − bi are not associates in J[i]. Finally, we
claim that this decomposition of p into the sum of squares of integers is unique
upto signs and the order in which a, b appear. Suppose decomposition is not
unique, therefore p = c2 + d2 = (c + di)(c − di). But then we know J[i] is a
Unique Factorization Domain, so c + di = u(a + ib), or c + di = u(a − bi), where
u is the unit element of J[i]. Taking values of u equal to 1, −1, i, −i one-by-one,
one can see either a = ±c or a = ±d. So decomposition of p into a2 + b2 is
unique upto sign and the order. Thus we concluded, when p is a prime of form
4n + 1, there corresponds exactly two prime elements(unique upto associates)
(a + bi), (a − bi) ∈ J[i], such that (a + bi) | p and (a − bi) | p.

Case 2 p is a prime of form 4n + 3: We claim p is prime in J[i] too. Suppose

p is composite in J[i], therefore p = z1 z2 with d(z1 ) 6= 1 and d(z2 ) 6= 1. Let
z1 = a + bi. So p = z1 z2 ⇒ d(p) = d(z1 z2 ) ⇒ p2 = d(z1 )d(z2 ) ⇒ p = d(z1 ) =
d(z2 ) ⇒ p = a2 + b2 . But by Problem 5, p of form 4n + 3 cannot be written as
sum of two squares, hence contradiction. So p is a prime element of J[i]. Thus
we concluded, p is the only prime element in J[i] such that p | p.

Case 3 when p = 2: This case is trivial to check 2 = (1 + i)(1 − i) with 1 + i

as prime element in J[i]. Note that 1 − i is associate of 1 + i. Thus 1 + i is the
only prime in J[i] dividing 2.

We summarize our finding that if z is a prime element in J[i], then either

z = a + bi or z = a − bi with d(z) = p where p is a prime of form 4n + 1 in Z;
or z = p where p is a prime of form 4n + 3 in Z; or z = 1 + i.
∗
9. Determine all positive integers which can be written as a sum of two squares
(of integers).
Solution: Clearly, n = 1 = 12 + 02 is expressible as sum of two squares. For
n 6= 1, we claim that n = p1 p2 · · · pl is expressible as sum of two squares if and
only if every prime factor pi of form 4k + 3 has even multiplicity. First suppose
it is given that n = p1 p2 · · · pl is expressible as sum of two squares, we need to
show that every prime factor pi of form 4k + 3 has even multiplicity. To prove it
by induction, we need to modify our statement to be proved. We assert that for
all n ∈ N, n 6= 1, we have either n not expressible as sum of two squares or if it
does then its every prime factor of form 4k + 3 has even multiplicity. Note that
we have excluded n = 1. When n = 2, we have 2 = 12 + 12 and 2 = 2, having no
prime factor of form 4k + 3. So the result is vacuously valid for n = 2. Suppose
the statement is valid for n = m − 1. We need to show that it is equally valid
for n = m. Now either m cannot be written as sum of two squares or it can be.
If it cannot be, then we have nothing left to prove. So we assume m = a2 + b2 .
Let m = p1 p2 · · · pl . Again if m has no prime factor of form 4k + 3, we have
nothing to prove. But suppose some prime factor pi0 of m is of form 4k + 3,
then we have pi0 | n ⇒ pi0 | (a2 + b2 ). Now we will work in J[i] to conclude
that pi0 | (a2 + b2 ) ⇒ pi | a and pi | b. Since pi0 is of form 4k + 3 so pi0 is also a
prime element in J[i]. But pi0 | a2 + b2 ⇒ pi0 | (a + bi)(a − bi) ⇒ pi0 | (a + bi)
or pi0 | (a − bi). When pi0 | (a + bi), we have (a + bi) = (c + di)pi0 for some
c + di ∈ J[i]. But that means a = cpi0 and b = dpi0 , or pi0 | a and pi0 | b.
Similarly, when pi0 | (a − bi), we have pi0 | a and pi0 | b. We return back to work
in Z. We have pi0 | a and pi0 | b, therefore a = pi0 a0 and b = pi0 b0 for some
a0 , b0 ∈ Z. So
n = p2i0 (a02 + b02 ) (1)
From (1) we can conclude n ≥ p2i0 . So either n = p2i0 or n > p2i0 . If n = p2 ,
our statement is validated for n = m. But if n > p2i0 , we define n0 = a02 + b02 .
Using (1), we have n0 = pn2 . But n > p2i0 , so n0 > 1. Also p2i0 > 1, so n0 < n.
i0

Invoking inductive hypothesis, we have n0 either not expressible as a sum of two

squares or if it does then its every prime factor of form 4k + 3 has even multi-
plicity. But then n = p2i0 n0 also is either not expressible as a sum of two squares
or if it does then its every prime factor of form 4k + 3 has even multiplicity.
Thus our assertion is true for n = m too. Thus our assertion is valid in general.
From our assertion(or modified statement), we conclude that if n is equal to sum
of two squares, then its every prime factor is of form 4k+3 is of even multiplicity.

Conversely, suppose it is given that n = p1 p2 · · · pl with every prime factor pi

of form 4k + 3 having even multiplicity, then we need to show n is expressible
as sum of two squares. Consider any pi , prime factor of n. Since pi is prime,
so either it is 2, or is of form 4k + 1, or is of form 4k + 3. When pi = 2, then
pi = 12 + 12 . When pi is of form 4k + 1, then pi = a2 + b2 for some a, b ∈ Z.
When pi is of form 4k + 3, then it is given to be of even multiplicity, therefore
p2j 2j
i for some j ≥ 1 must be the factor of n. We can treat pi = (pi ) + 0 .
j 2 2

Thus, we saw n is a product of sum of two squares. Also we observe that

(a2 + b2 )(c2 + d2 ) = (ac − bd)2 + (ad + bc)2 , i.e. product of two numbers which
are expressible as sum of two squares is again a number which can be expressed
as sum of two squares. One can apply this observation over and over again to
see product of sum of two squares is again a sum of two squares. Hence n is
a product of two squares. Thus n with its every prime factor of form 4k + 1
having even multiplicity implies n is expressible as sum of two squares.

Thus n = p1 p2 · · · pl with n 6= 1 is expressible as sum of two squares if and only

if every prime factor pi of form 4k + 3 has even multiplicity. With this result
at our disposal, we can determine all integers which can be expressed as sum of
two squares.
Problems (Page 158)
1. Find the greatest common divisor of the following polynomials over F , the
field of rational numbers:
(a) x3 − 6x2 + x + 4 and x5 − 6x + 1.
(b) x2 + 1 and x6 + x3 + x + 1.
Solution:
(a) Using long division method, we have

x5 − 6x + 1 = (x3 − 6x2 + x + 4)(x2 + 6x + 35) + 200x2 − 65x − 139

So gcd(x5 − 6x + 1, x3 − 6x2 + x + 4) = gcd(x3 − 6x2 + x + 4, 200x2 − 65x − 139).

Again we have
x 227 239 447
x3 − 6x2 + x + 4 = (200x2 − 65x − 139)( − )− x+
200 8000 1600 8000
239 447


So gcd(x3 −6x2 +x+4, 200x2 −65x−139) = gcd 200x2 − 65x − 139, − 1600 x + 8000 .
Again we have
  
2 239 447 320000 3752000 7730176
200x − 65x − 139 = − x+ − x− −
1600 8000 239 57121 57121
239 447 239 447
, − 7730176



So gcd 200x2 − 65x − 139, − 1600 x + 8000 = gcd − 1600 x + 8000 57121 =
1. Hence
gcd(x5 − 6x + 1, x3 − 6x2 + x + 4) = 1
(b) Using long division method, we have

x6 + x3 + x + 1 = (x2 + 1)(x4 − x2 + x + 1)

So we have gcd(x6 + x3 + x + 1, x2 + 1) = gcd(x2 + 1, 0) = x2 + 1. So we have

gcd(x6 + x3 + x + 1, x2 + 1) = x2 + 1

2. Prove that
(a) x2 + x + 1 is irreducible over F, the field of integers mod 2.
(b) x2 + 1 is irreducible over the integers mod 7.
(c) x3 − 9 is irreducible over the integers mod 31.
(d) x3 − 9 is reducible over the integers mod 11.
Solution:
(a) We have

x2 + x + 1 |x=0 = 1 mod 2
x2 + x + 1 |x=1 = 1 mod 2

So x2 + x + 1 6= 0 ∀ x ∈ Z2 , implying x2 + x + 1 is irreducible in Z2 [x].

(b) We have

x2 + 1 |x=0 = 1 mod 7
x2 + 1 |x=1 = 2 mod 7
x2 + 1 |x=2 = 5 mod 7
x2 + 1 |x=3 = 3 mod 7
x2 + 1 |x=4 = 3 mod 7
x2 + 1 |x=5 = 5 mod 7
x2 + 1 |x=6 = 2 mod 7

So x2 + 1 6= 0 ∀ x ∈ Z7 . So x2 + 1 is irreducible in Z7 [x]

(c) We have

x3 − 9 |x=0 = 22 mod 31
x3 − 9 |x=1 = 23 mod 31
x3 − 9 |x=2 = 30 mod 31
x3 − 9 |x=3 = 18 mod 31
x3 − 9 |x=4 = 24 mod 31
x3 − 9 |x=5 = 23 mod 31
x3 − 9 |x=6 = 21 mod 31
x3 − 9 |x=7 = 24 mod 31
x3 − 9 |x=8 = 7 mod 31
x3 − 9 |x=9 = 7 mod 31
x3 − 9 |x=10 = 30 mod 31
x3 − 9 |x=11 = 20 mod 31
x3 − 9 |x=12 = 14 mod 31
x3 − 9 |x=13 = 18 mod 31
x3 − 9 |x=14 = 7 mod 31
x3 − 9 |x=15 = 18 mod 31
x3 − 9 |x=16 = 26 mod 31
x3 − 9 |x=17 = 6 mod 31
x3 − 9 |x=18 = 26 mod 31
x3 − 9 |x=19 = 30 mod 31
x3 − 9 |x=20 = 24 mod 31
x3 − 9 |x=21 = 14 mod 31
x3 − 9 |x=22 = 6 mod 31
 x3 − 9 |x=23 = 6 mod 31
x3 − 9 |x=24 = 20 mod 31
x3 − 9 |x=25 = 23 mod 31
x3 − 9 |x=26 = 21 mod 31
x3 − 9 |x=27 = 20 mod 31
x3 − 9 |x=28 = 26 mod 31
x3 − 9 |x=29 = 14 mod 31
x3 − 9 |x=30 = 21 mod 31

So x3 − 9 6= 0 ∀ x ∈ Z31 . So x3 − 9 is irreducible in Z31 [x]

(d) We have

x3 − 9 |x=0 = 1 mod 11
x3 − 9 |x=1 = 3 mod 11
x3 − 9 |x=2 = 10 mod 11
x3 − 9 |x=3 = 7 mod 11
x3 − 9 |x=4 = 0 mod 11
x3 − 9 |x=5 = 5 mod 11
x3 − 9 |x=6 = 9 mod 11
x3 − 9 |x=7 = 4 mod 11
x3 − 9 |x=8 = 8 mod 11
x3 − 9 |x=9 = 5 mod 11
x3 − 9 |x=10 = 1 mod 11

So x3 − 9 = 0 for x = 4. Therefore x − 4, or x + 7 is a factor. We can see by

long division x3 − 9 = (x + 7)(x2 + 4x + 5). So x3 − 9 is reducible in Z11 [x].

3. Let F, K be two fields F ⊂ K and suppose f (x), g(x) ∈ F [x] are relatively
prime in F [x]. Prove that they are relatively prime in K[x].
Solution: First we can easily see that if 1 is the multiplicative identity of F ,
then it is also the multiplicative identity of K too. Now since f (x), g(x) are rel-
atively prime in F [x], so 1 = λ(x)f (x) + µ(x)g(x), for some λ(x), µ(x) ∈ F [x].
But since F ⊂ K, therefore 1, λ(x), µ(x), f (x), g(x) are also elements of K[x].
So the relation 1 = λ(x)f (x) + µ(x)g(x) is equally valid in K[x]. But that
would mean f (x), g(x) as elements of K[x] are relatively prime in K[x]. Hence
the result.
4. (a) Prove that x2 + 1 is irreducible over the field F of integers mod 11 and
prove directly that F [x]/(x2 + 1) is a field having 121 elements.
(b) Prove that x2 + x + 4 is irreducible over F , the field of integers mod 11 and
prove directly that F [x]/(x2 + x + 4) is a field having 121 elements.
∗
(c) Prove that the fields of part (a) and (b) are isomorphic.
Solution:
(a) We have
x2 + 1 |x=0 = 1 mod 11
x2 + 1 |x=1 = 2 mod 11
x2 + 1 |x=2 = 5 mod 11
x2 + 1 |x=3 = 10 mod 11
x2 + 1 |x=4 = 6 mod 11
x2 + 1 |x=5 = 4 mod 11
x2 + 1 |x=6 = 4 mod 11
x2 + 1 |x=7 = 6 mod 11
x2 + 1 |x=8 = 10 mod 11
x2 + 1 |x=9 = 5 mod 11
x2 + 1 |x=10 = 2 mod 11
So x2 + 1 6= 0 ∀ x ∈ F . So x2 + 1 is irreducible over F [x].

Now consider F [x]/hx2 + 1i. Since hx2 + 1i is an ideal of F [x], so F [x]/hx2 + 1i

is a ring. Also F [x]/hx2 + 1i = {hx2 + 1i + ax + b | a, b ∈ F }. Since F has 11
elements, so F [x]/hx2 + 1i has 11 × 11 = 121 elements. F being commutative
implies F [x]/hx2 + 1i is also commutative. Next we will prove F [x]/hx2 + 1i
to be an integral domain, which would, in turn will prove F [x]/hx2 + 1i to
be a field as every finite integral domain is a field. Suppose (hx2 + 1i + ax +
b)(hx2 + 1i + cx + d) = hx2 + 1i, with (hx2 + 1i + ax + b) 6= hx2 + 1i. But
(hx2 + 1i + ax + b)(hx2 + 1i + cx + d) = hx2 + 1i implies
hx2 + 1i + (ax + b)(cx + d) = hx2 + 1i
⇒ hx2 + 1i + acx2 + (ad + bc)x + bd = hx2 + 1i
⇒ hx2 + 1i + ac(x2 + 1) + (ad + bc)x + (bd − ac) = hx2 + 1i
⇒ hx2 + 1i + (ad + bc)x + (bd − ac) = hx2 + 1i
⇒ (ad + bc)x + (bd − ac) ∈ hx2 + 1i
⇒ ad + bc = bd − ac = 0 mod 11 (1)
(hx2 + 1i + ax + b) 6= hx2 + 1i implies a = b 6= 0 mod 11 simultaneously. Suppose
a = 0 mod 11, therefore b 6= 0 mod 11. In this case, (1) reduces to
bc = 0 mod 11
(2)
bd = 0 mod 11
But since b 6= 0 mod 11 and F being a field, so (2) implies c = d = 0 mod 11.
Similarly, if b = 0 mod 11, then also c = d = 0 mod 11. Next if both a, b 6= 0
mod 11, then (1) reduces to

(a2 + b2 )c = 0 mod 11
(3)
(a2 + b2 )d = 0 mod 11

Let 11 = p. Suppose, if possible a2 + b2 = 0 mod p. So a2 + b2 = kp for some

positive integer k. Note that p is a prime of form 4n + 3. So using Problem
9 (page 152 of the book) we have pi with i a positive odd integer as a factor
of k. Thus at least p | k. So at least p2 | a2 + b2 . Working in J[i], we have
a2 + b2 = (a + bi)(a − bi). So p2 | (a + bi)(a − bi), which gives three possibilities,
1) p | (a + bi); 2) p2 | (a + bi); 3) p2 | (a − bi). All three possibilities, imply p | a
and p | b (Why). So a = b = 0 mod 11. Thus a2 + b2 = 0 mod 11 implies
a = b = 0 mod 11. So when a, b ∈ F − {0}, we have a2 + b2 6= 0 mod 11. So
c = d = 0 mod 11. Thus we see (hx2 + 1i + ax + b)(hx2 + 1i + cx + d) = hx2 + 1i,
with (hx2 + 1i + ax + b) 6= hx2 + 1i implies hx2 + 1i + cx + d = hx2 + 1i. Hence
F [x]/hx2 + 1i is an integral domain. Ans so, being finite, it is a field.

(b) We have

x2 + x + 4 |x=0 = 4 mod 11
x2 + x + 4 |x=1 = 6 mod 11
x2 + x + 4 |x=2 = 10 mod 11
x2 + x + 4 |x=3 = 5 mod 11
x2 + x + 4 |x=4 = 2 mod 11
x2 + x + 4 |x=5 = 1 mod 11
x2 + x + 4 |x=6 = 2 mod 11
x2 + x + 4 |x=7 = 5 mod 11
x2 + x + 4 |x=8 = 10 mod 11
x2 + x + 4 |x=9 = 6 mod 11
x2 + x + 4 |x=10 = 4 mod 11

So x2 + x + 4 6= 0 ∀ x ∈ F . So x2 + x + 4 is irreducible over F [x].

Again as we see in part (a), F [x]/hx2 + x + 4i is a commutative ring with 121

elements. To show F [x]/hx2 +x+4i a field we will first prove it to be an integral
domain. Suppose (hx2 + x + 4i + ax + b)(hx2 + x + 4i + cx + d) = hx2 + x + 4i
with hx2 + x + 4i + ax + b 6= hx2 + x + 4i. So we have

hx2 + x + 4i + (ax + b)(cx + d) = hx2 + x + 4i

⇒ hx2 + x + 4i + acx2 + (ad + bc)x + bd = hx2 + x + 4i
⇒ hx2 + x + 4i + ac(x2 + x + 4) + (ad + bc − ac)x + (bd − 4ac) = hx2 + x + 4i
⇒ (ad + bc − ac)x + (bd − 4ac) ∈ hx2 + x + 4i
⇒ ad + bc − ac = bd − 4ac = 0 mod 11 (1)

If a = 0 mod 11, then b 6= 0 mod 11 as hx2 + x + 4i + ax + b 6= hx2 + x + 4i.

But then (1) reduces to
bc = 0 mod 11
(2)
bd = 0 mod 11
Since b 6= 0 mod 11, therefore c = d = 0 mod 11. Similarly if b = 0 mod 11, we
have a 6= 0 mod 11, and (1) reduces to

a(d − c) = 0 mod 11
(3)
−4ac = 0 mod 11

But a 6= 0 mod 11, so c = 0 mod 11, which in turn forces d = 0 mod 11. Thus
c = d = 0 in this case too. Finally suppose both a, b 6= 0 mod 11, then (1)
reduces to
(4a2 + b2 − ba)c = 0 mod 11
(4)
(4a2 + b2 − ba)d = 0 mod 11
Now 4a2 +b2 −ab = 4a2 +b2 +10ab = 25a2 +b2 +10ab−21a2 = (5a+b)2 +a2 (We
are working in modulo 11). As in previous part if 4a2 +b2 −ab = (5a+b)2 +a2 = 0
mod 11, then 11 | a and 11 | (5a + b). Thus 4a2 + b2 − ab = 0 mod 11 implies
a = b = 0 mod 11. In other words if a, b 6= 0 mod 11, then 4a2 + b2 − ab 6= 0
mod 11. But then F being a field makes (4) implying c = d = 0 mod 11.
Thus (hx2 + x + 4i + ax + b)(hx2 + x + 4i + cx + d) = hx2 + x + 4i with
hx2 + x + 4i + ax + b 6= hx2 + x + 4i implies hx2 + x + 4i + cx + d = hx2 + x + 4i.
So F [x]/hx2 + x + 4i is an integral domain, and hence is a field.

(c) Define φ : F [x]/hx2 +1i −→ F [x]/hx2 +x+4i such that φ(hx2 +1i+ax+b) =
hx2 + x + 4i + ax + (b + 6a). We claim φ is an one-to-one and onto ring
homomorphism. Firstly, one can easily see that mapping is well-defined. Next
we prove mapping is a ring homomorphism. We have

φ((hx2 + 1i + ax + b) + (hx2 + 1i + a0 x + b0 ))
= φ((hx2 + 1i + (a + a0 )x + (b + b0 ))
= hx2 + x + 4i + (a + a0 )x + (b + b0 ) + 6(a + a0 )
= hx2 + x + 4i + (ax + b + 6a) + (a0 x + b0 + 6a0 )
= (hx2 + x + 4i + (ax + b + 6a)) + (hx2 + x + 4i + (a0 x + b0 + 6a0 ))
= φ(hx2 + 1i + ax + b) + φ(hx2 + 1i + a0 x + b0 )

Also we have

φ((hx2 + 1i + ax + b)(hx2 + 1i + a0 x + b0 ))
= φ(hx2 + 1i + aa0 x2 + (ab0 + ba0 )x + bb0 )
 = φ(hx2 + 1i + aa0 (x2 + 1) + (ab0 + ba0 )x + (bb0 − aa0 ))
= φ(hx2 + 1i + (ab0 + ba0 )x + (bb0 − aa0 ))
= hx2 + x + 4i + (ab0 + ba0 )x + (bb0 − aa0 ) + 6(ab0 + ba0 )
= hx2 + x + 4i + (ab0 + ba0 )x + (bb0 + 6ab0 + 6ba0 − aa0 ) (1)

Also

φ(hx2 + 1i + ax + b)φ(hx2 + 1i + a0 x + b0 )
= (hx2 + x + 4i + ax + (b + 6a))(hx2 + x + 4i + a0 x + (b0 + 6a0 ))
= hx2 + x + 4i + aa0 x2 + (a(b0 + 6a0 ) + (b + 6a)a0 )x+
(b + 6a)(b0 + 6a0 )
= hx2 + x + 4i + aa0 (x2 + x + 4) + (a(b0 + 6a0 ) + (b + 6a)a0 − aa0 )x+
(b + 6a)(b0 + 6a0 ) − 4aa0
= hx2 + x + 4i + (ab0 + 6aa0 + ba0 + 6aa0 − aa0 )x+
(bb0 + 6ab0 + 6ba0 + 36aa0 − 4aa0 )
= hx2 + x + 4i + (ab0 + ba0 )x + (bb0 + 6ab0 + 6ba0 − aa0 ) (2)

So from (1) and (2) we have φ((hx2 + 1i + ax + b)(hx2 + 1i + a0 x + b0 )) =

φ(hx2 + 1i + ax + b)φ(hx2 + 1i + a0 x + b0 ) Hence φ is a ring homomorphism. Next
we prove mapping φ is one-to-one mapping. We have

φ(hx2 + 1i + ax + b) = φ(hx2 + 1i + a0 x + b0 )
⇒ hx2 + x + 4i + ax + b + 6a = hx2 + x + 4i + a0 x + b0 + 6a0
⇒ (a − a0 )x + (b + 6a − b0 − 6a0 ) ∈ hx2 + x + 4i
⇒ (a − a0 )x + (b + 6a − b0 − 6a0 ) = 0 mod 11
⇒ (a − a0 ) = (b + 6a − b0 − 6a0 ) = 0 mod 11
⇒ a = a0 mod 11 and b = b0 mod 11
⇒ hx2 + 1i + ax + b = hx2 + 1i + a0 x + b0

Hence φ is an one-to-one mapping. Also if some ỹ = hx2 + x + 4i + ax + b ∈

F [x]/hx2 +x+4i, then x̃ = hx2 +1i+ax+(b+5a) ∈ F [x]/hx2 +1i is the inverse-
image of ỹ. Thus inverse-image of every ỹ ∈ F [x]/hx2 + x + 4i exists. So φ is
onto too. Thus we have mapping φ an one-to-one and onto ring homomorphism.
So
F [x] F [x]
2
≈ 2
hx + 1i hx + x + 4i

5. Let F be the field of real numbers. Prove that F [x]/(x2 + 1) is a field

isomorphic to the field of complex numbers.
Solution: We have x2 + 1 a irreducible element of F [x], where F is the field of
real numbers. Therefore hx2 + 1i is a maximal ideal of F [x]. So F [x]/hx2 + 1i
is field. Moreover F [x]/hx2 + 1i = {hx2 + 1i + ax + b | a, b ∈ F }. To exhibit an
one-to-one and onto homomorphism, we define mapping φ : F [x]/hx2 + 1i −→ C
such that φ(hx2 + 1i + ax + b) = b + ia, where C is the field of complex numbers.
Clearly mapping φ is well-defined. Also mapping is one-to-one and onto too.
Next

φ((hx2 + 1i + ax + b) + (hx2 + 1i + a0 x + b0 ))
= φ(hx2 + 1i + (a + a0 )x + (b + b0 ))
= b + b0 + i(a + a0 )
= (b + ia) + (b0 + ia0 )
= φ(hx2 + 1i + ax + b) + φ(hx2 + 1i + a0 x + b0 )

and

φ((hx2 + 1i + ax + b)(hx2 + 1i + a0 x + b0 ))
= φ(h+x2 + 1i + (ax + b)(a0 x + b0 ))
= φ(hx2 + 1i + aa0 x2 + (ab0 + ba0 )x + bb0 )
= φ(hx2 + 1i + aa0 (x2 + 1) + (ab0 + ba0 )x + bb0 − aa0 )
= φ(hx2 + 1i + (ab0 + ba0 )x + (bb0 − aa0 ))
= (bb0 − aa0 ) + i(ab0 + ba0 )
= (a + ib)(a0 + ib0 )
= φ(hx2 + 1i + ax + b)φ(hx2 + 1i + a0 x + b0 )

Thus mapping φ is a ring homomorphism too. Hence F [x]/hx2 + 1i ≈ C.

∗
6. Define the derivative f 0 (x) of the polynomial

f (x) = a0 + a1 x + · · · + an xn
as f 0 (x) = a1 + 2a2 x + 3a3 x2 + · · · + nan xn−1 .

Prove that if f (x) ∈ F [x], where F is the field of rational numbers, then f (x)
is divisible by the square of a polynomial if and only if f (x) and f 0 (x) have a
greatest common divisor d(x) of positive degree.
Solution: We first assert two results which we are going to use in the proof.
1. If f (x) = g(x)h(x), then f 0 (x) = g 0 (x)h(x) + g(x)h0 (x);
2. If f (x) = (g(x))n for some n ∈ N, then f 0 (x) = n(g(x))n−1 g 0 (x).
We left the proof of above results as an exercise for the reader.

Now first suppose, f (x) is divisible by the square of a polynomial, say h(x) with
deg(h(x)) ≥ 1. Therefore, f (x) = (h(x))2 g(x). So f 0 (x) = (2h(x)h0 (x))g(x) +
(h(x))2 g 0 (x) = h(x)(2h0 (x)g(x) + h(x)g 0 (x)). But that means h(x) | f (x) and
h(x) | f 0 (x). So h(x) | gcd(f (x), f 0 (x)), or gcd(f (x), f 0 (x)) = h(x)q(x) for some
q(x) ∈ F [x]. Also deg(h(x)q(x)) = deg(h(x)) + deg(q(x)) ≥ 1 as deg(h(x)) ≥ 1.
Thus greatest common divisor of f (x) and f 0 (x) is of positive degree.

Conversely, suppose some gcd(f (x), f 0 (x)) = h(x) with deg(h(x)) ≥ 1. So

f (x) = h(x)g(x) for some g(x) ∈ F [x]. Also f 0 (x) = h0 (x)g(x)+h(x)g 0 (x), there-
fore h(x) | h0 (x)g(x)+h(x)g 0 (x) ⇒ h(x) | h0 (x)g(x). But deg(h0 (x)) < deg(h(x)),
so some irreducible factor p(x) of h(x) does not divide h0 (x), because if every ir-
reducible factor of h(x) divides h0 (x) would imply deg(h0 (x)) ≥ deg(h(x)) which
is not the case. So some p(x) | h(x) and p(x)6 | h0 (x). But h(x) | h0 (x)g(x),
therefore p(x) | h0 (x)g(x). So p(x) | g(x) as p(x)6 | h0 (x). But p(x) | h(x) and
p(x) | g(x) implies (p(x))2 | h(x)g(x). Thus (p(x))2 | f (x). Also p(x) being irre-
ducible implies deg(p(x)) ≥ 1. Thus we concluded, some f (x) ∈ F [x] is divisible
by the square of a polynomial with positive degree if and only if greatest com-
mon divisor of f (x) and f 0 (x) is of positive degree.

7. If f (x) is in F [x], where F is the field of integers mod p, p a prime, and f (x)
irreducible over F of degree n prove that F [x]/(f (x)) is a field with pn elements.
Solution: Since f (x) is irreducible, therefore by Lemma 3.9.6 hf (x)i is a
maximal ideal of F [x]. Then using Theorem 3.5.1, we have F [x]/hf (x)i is
a field. Also every element of F [x]/hf (x)i can be uniquely represented as
hf (x)i + an−1 xn−1 + · · · + a1 x + a0 with ai ∈ F , so the field F [x]/hf (x)i has pn
elements. Hence the result.
Problems (Page 161)
1. Let D be a Euclidean ring, F its field of quotients. Prove the Gauss Lemma
for polynomials with coefficients in D factored as products of polynomials with
coefficients in F .
Solution: Suppose some f (x) ∈ D[x]. Therefore f (x) ∈ F [x] too. Also suppose
f (x) = g(x)h(x) for some g(x), h(x) ∈ F [x]. But then using Problem 11 (Page
0
166 of the book), we have g(x) = g λ(x) for some g 0 (x) ∈ D[x] and λ ∈ D. Also
if content of g 0 (x) is d1 , then g 0 (x) = d1 g 00 x, where g 00 (x) ∈ D[x] is a primitive.
So g(x) = dλ1 g 00 (x). Similarly, h(x) = dµ2 h00 (x) where h00 (x) is primitive in D[x]
and d2 , µ ∈ D. So we have
d1 d2 00
f (x) = g (x)h00 (x)
λµ
Also content of g 00 (x), h00 (x) equal to 1 implies content of g 00 (x)h00 (x) is also 1.
Now f (x), g 00 (x)h00 (x) ∈ D[x] with content 1, therefore dλµ1 d2
= 1. So we have
00 00
f (x) = g (x)h (x) showing f (x) is reducible in D[x] too. Thus if f (x) ∈ D[x]
is reducible in F [x], then f (x) is also reducible in D[x]. Hence the result.

2. If p is a prime number, prove that the polynomial xn − p is irreducible over

the rationals.
Solution: Let f (x) = xn − p = an xn + an−1 xn−1 + · · · + a0 . So an = 1, a0 = −p
and rest ai = 0. We apply Eisenstein criterion. We have f (x) ∈ Z[x], and
p | a0 , p | a1 , · · · , p | an−1 , and p6 | an . Also p2 6 | a0 . Therefore f (x) is irreducible
in Q[x]. Hence the result.

3. Prove that the polynomial 1 + x + · · · + xp−1 , where p is a prime number, is

irreducible over the field of rational numbers. (Hint: Consider the polynomial
1 + (x + 1) + (x + 1)2 + · · · + (x + 1)p−1 , and use the Eisenstein criterion.)
Solution: Let f (x) = 1 + x + · · · + xp−1 . Now f (x) is irreducible in Q[x] if and
only if f (x + 1) is irreducible in Q[x]. We have xp − 1 = (x − 1)(xp−1 + xp−2 +
p
−1
· · · + 1), therefore f (x) = xx−1 (undefined notation?). So we have

(x + 1)p − 1
f (x + 1) =
(x + 1) − 1
1 n
= ( C1 x + n C2 x2 + · · · + n Cp xp )
x
= n C1 + n C2 x + · · · + n Cp−1 xp−2 + n Cp xp−1

Now we have n Cr r! = p(p − 1) · · · (p − r + 1) for r > 0. Also p6 | r! for r < p as p

is prime. So if 1 ≤ r ≤ p − 1, we have with p6 | r! and p | p(p − 1) · · · (p − r + 1),
implying p | n Cr . Thus p divides all coefficient of f (x + 1) except for the coeffi-
cient of xp−1 which is 1. Also p2 6 | p, the constant coefficient of f (x + 1). So by
Eisenstein criterion f (x+ 1) is irreducible in Q[x]. And hence f (x) is irreducible
in Q.

4. If m and n are relatively prime integers and if

 m
x− | (a0 + a1 x + · · · + ar xr ),
n
where the a’s are integers, prove that m | a0 and n | ar .
Solution: Let

a0 + a1 x + · · · + ar xr = (x − m/n)(b0 + b1 x + · · · + br−1 xr−1 ) (1)

Comparing coefficients of x0 , x1 , x2 , · · · and expressing in terms of ai , we have

n
b0 = − a0
m
n n 
b1 = − a1 + a0
m m
n n n 
b2 = − a2 + a1 + a0
m m m
..
.
n  n 
br−1 = − ar−1 + (ar−2 + · · · )
m m
But we have br−1 = ar . Therefore
n  n 
ar = − ar−1 + (ar−2 + · · · )
m m
Also gcd(m, n) = 1, so n | ar . Again comparing coefficients of xr , xr−1 , · · · in
(1) and expressing in terms of ai , we have

br−1 = ar
m
br−2 = ar−1 + ar
n 
m m 
br−3 = ar−2 + ar−1 + ar
n n
..
.
m m 
b0 = a 1 + a2 + (· · · )
n n
n
But we have b0 = − m a0 . Therefore
n m m 
− a0 = a1 + a2 + (· · · )
m n n
or
m m m 
a0 = − a1 + a2 + (· · · )
n n n
Since gcd(m, n) = 1, therefore m | a0 . Hence the result.

5. If a is rational and x − a divides an integer monic polynomial, prove that a

must be an integer.
Solution: Suppose a is a rational number, therefore we can assume a = pq
with gcd(p, q) = 1. Let f (x) ∈ Z[x] be some monic polynomial. Let f (x) =
xm + am−1 xm−1 + · · · + a0 for some m ∈ N. We are give (x − p/q) | f (x).
So f (x) = (x − p/q)g(x) for some g(x) ∈ Q[x]. Since g(x) ∈ Q[x], therefore
g(x) = λd g 0 (x), where g 0 (x) is primitive in Z[x] and d, λ ∈ Z. So we have
 
d p
f (x) = x− g 0 (x)
λ q
d
= (qx − p)g 0 (x) (1)
λq

Now since gcd(p, q) = 1, therefore qx − p is a primitive in Z[x]. Also g 0 (x) is

primitive in Z[x], so (qx − p)g 0 (x) is primitive in Z[x]. Also f (x) being monic,
d
therefore is primitive in Z[x]. So from equation (1) we have λq = 1. So we have

f (x) = (qx − p)g 0 (x) (2)

Since g 0 (x) ∈ Z[x], therefore let g 0 (x) = b0 + b1 x + · · · + bm−1 xm−1 with all
bi ∈ Z. Comparing coefficient of xm in equation (2), we have 1 = qbm−1 . But
that means q is a unit in Z, or q = ±1, showing p/q is an integer. Hence a = p/q
is an integer.
Problems (Page 166)
1. Prove that R[x] is a commutative ring with unit element whenever R is.
Solution: We are given R is a commutative ring with unity element. Suppose
some f (x) = am xm + · · · + a0 and g(x) = bn xn + · · · + b0 are elements in R[x].
Let f (x)g(x) = cm+n xm+n + · · · + c0 and g(x)f (x) = dm+n xm+n + · · · + d0 . So
we have for 0 6 i 6 m + n
i
X
ci = aj bi−j
j=0
i
X
= bi−j aj as R is commutative
j=0
0
X
= bk ai−k
k=i
i
X
= bk ai−k
k=0
= di

So f (x)g(x) = g(x)f (x), implying R[x] is commutative too. Also if 1 is the mul-
tiplicative identity of R, we have f (x)1 = 1f (x) = f (x). Thus multiplicative
identity of R is also the multiplicative identity of R[x]. And hence R[x] too is
a commutative ring and has unity element.

2. Prove that R[x1 , . . . , xn ] = R[xi1 , . . . , xin ], where (i1 , . . . , in ) is a permutation

of (1, 2, . . . , n).
Solution: Let some f (x1 , · · · , xn ) ∈ R[x1 , . . . , xn ]. Therefore
X
f (x1 , · · · , xn ) = aj1 ,j2 ,··· ,jn xj11 xj22 · · · xjnn , where aj1 ,j2 ,··· ,jn ∈ R

j j j
Also xj11 xj22 · · · xjnn = xi1i1 xi2i2 · · · xinin , so we have
X
f (x1 , · · · , xn ) = aj1 ,j2 ,··· ,jn xj11 xj22 · · · xjnn
j j j
X
= aj1 ,j2 ,··· ,jn xi1i1 xi2i2 · · · xinin ∈ R[xi1 , . . . , xin ]

So R[x1 , . . . , xn ] ⊂ R[xi1 , . . . , xin ]. Similarly, we can show R[xi1 , . . . , xin ] ⊂

R[x1 , . . . , xn ]. Hence R[x1 , . . . , xn ] = R[xi1 , . . . , xin ]

3. If R is an integral domain, prove that for f (x), g(x) in R[x], deg(f (x)g(x)) =
deg(f (x)) + deg(g(x)).
Solution: Let f (x) = a0 + a1 x + · · · , with deg(f (x)) = m for some m ∈ N.
So we have am 6= 0 and ai = 0 ∀ i > m. Also let g(x) = b0 + b1 x + · · · ,
with deg(g(x)) = n for some n ∈ N. So bn 6= 0 and bi = 0 ∀ i > n. Let
h(x) = f (x)g(x) = c0 + c1 x + · · · . We have
m+n
X
cm+n = ai bm+n−i
i=0
= (a0 bm+n + · · · + am−1 bn+1 ) + am bn + (am+1 bn−1 + · · · + am+n b0 )
= 0 + am bn + 0
6= 0 as R is an integral domain and am 6= 0 and bn 6= 0

Also for some integer k > 1 we have

m+n+k
X
cm+n+k = ai bm+n−i
i=0
= (a0 bm+n + · · · + am−1 bn+k+1 + am bn+k )+
(am+1 bn+k−1 + · · · + am+n+k b0 )
=0+0=0

So deg(h(x)) = m + n, or deg(f (x)g(x)) = deg(f (x)) + deg(g(x)). Hence the

result.

4. If R is an integral domain with unit element, prove that any unit in R[x]
must already be a unit in R.
Solution: Suppose f (x) be some unit in R[x], therefore there exists some
g(x) ∈ R[x] such that f (x)g(x) = 1, where 1 is the multiplicative identity of R.
Now we have deg(1) = deg(f (x)g(x)) = deg(f (x)) + deg(g(x)). But deg(1) = 0,
therefore deg(f (x)) + deg(g(x)) = 0. Also deg(f (x)), deg(g(x)) ≥ 0, therefore
we have only possibility that deg(f (x)) = deg(g(x)) = 0. But that means
f (x) = a0 and g(x) = b0 for some a0 , b0 ∈ R. So f (x)g(x) = 1 ⇒ a0 b0 = 1. But
that means a0 is a unit in R. Thus any unit of R[x] is a unit of R too. Hence
the result.

5. Let R be a commutative ring with no nonzero nilpotent elements (that is, an

implies a = 0). If f (x) = a0 + a1 x + · · · + am xm in R[x] is a zero-divisor, prove
that there is an element b 6= 0 in R such that ba0 = ba1 = · · · = bam = 0.
Solution: Since f (x) is the zero-divisor in R[x], therefore there exist g(x) ∈ R[x]
with g(x) 6= 0 such that f (x)g(x) = 0. Let deg(g(x)) = n. Thus g(x) =
b0 + b1 x + · · · + bn xn with bn 6= 0. Also for i ∈ N, ai = 0 ⇒ a = 0, therefore
bin 6= 0 for all i ∈ N as bn 6= 0. Let f (x)g(x) = c0 + c1 x + · · · + cm+n xm+n .
But f (x)g(x) = 0, therefore ci = 0 ∀ i. So cm+n = am bn = 0. We claim
am−j bj+1
n = 0 for all 0 ≤ j ≤ m and we establish our claim by induction
over j. As am bn = 0, so the result is true for the base case, j = 0. Suppose
am−j bj+1
n = 0 for all 0 ≤ j ≤ k − 1. We need to show, result is valid for j = k
too. We have

cm+n−k = am−k bn + am−k+1 bn−1 + · · · + am bn−k = 0

⇒ am−k bn bkn + am−k+1 bn−1 bkn + · · · + am bn−k bkn = 0bkn
⇒ am−k bk+1
n + 0 + ··· + 0 = 0
⇒ am−k bk+1
n =0 (1)

So the result holds good for j = k too. Thus we have am bn = am−1 b2n = · · · =
am−j bj+1
n = · · · = a0 bm+1
n = 0. But that also means am bm+1
n = am−1 bm+1
n =
m+1
· · · = a0 bn = 0, or am b = am−1 b = · · · = a0 b = 0, where b = bm+1
n . Hence
the result.

6. Do Problem 5 dropping the assumption that R has no nonzero nilpotent

elements.
Solution: We choose some g(x) ∈ R[x] from the set {gi (x) | f (x)gi (x) = 0}
such that deg(g(x)) ≤ deg(gi (x)) ∀i. Note that the existence of such g(x)
is guaranteed as S = {deg(gi (x)) | gi (x)f (x) = 0} is a non-empty set and is
bounded from below. Now let deg(g(x)) = n for some n ∈ N and therefore,
let g(x) = b0 + b1 x + · · · + bn xn with bn 6= 0. Also f (x)g(x) = 0. Comparing
coefficient of xm+n in f (x)g(x), we have am bn = 0. So am (f (x)g(x)) = am 0 ⇒
(am g(x))f (x) = 0. Let g 0 (x) = am g(x). So g 0 (x) is a polynomial of degree less
than n with g 0 (x)f (x) = 0, which is not possible as g(x) is the polynomial with
degree less than or equal to the degree of all polynomials with gi (x)f (x) = 0.
So
am g(x) = 0 (1)
We claim am−j g(x) = 0 for 0 ≤ j ≤ m. We establish our claim by induction
over j. Base case with j = 0 has already been shown holding true. Suppose
am−j g(x) = 0 for 0 ≤ j ≤ k − 1, i.e. result holds good upto j = k − 1. We need
to show result holds good for j = k too, i.e. am−k g(x) = 0. We have
k−1
X
f (x)g(x) − am−j xm−j g(x) = 0
j=0
k−1
X
⇒ (f (x) − am−j xm−j )g(x) = 0
j=0

⇒ (a0 + a1 x + · · · + an−k xn−k )g(x) = 0

⇒ am−k bn = 0 (2)

Now we have f (x)g(x) = 0 ⇒ am−k (g(x)f (x)) = am−k 0 ⇒ (am−k g(x))f (x) =
0. Again let am−k g(x) = g 0 (x), therefore g 0 (x)f (x) = 0. But (2) implies
deg(g 0 (x)) < deg(g(x)), which is not possible, forcing g 0 (x) = 0. So am−k g(x) =
0, showing result is valid for j = k too. Hence result is valid for all possible
j. Thus, we have ai g(x) = 0 ∀ i. In particular, ai bn = 0 for all i. So if
b = bn 6= 0, we have a0 b = a1 b = · · · = am b = 0. Hence the result.

∗
7. If R is a commutative ring with unit element, prove that a0 +a1 x+· · ·+an xn
in R[x] has an inverse in R[x] (i.e., is unit in R[x]) if and only if a0 is a unit in
R and a1 , . . . , an are nilpotent elements in R.
Solution: We first prove two results required to prove the main result. First
we claim that in a ring R̃, if x̃, ỹ are nilpotent, then so is x̃ + ỹ. Since x̃ and ỹ
are nilpotent, so x̃l = 0 and ỹ m = 0 for some l, m ∈ N. But then (x̃ + ỹ)l+m = 0,
showing x̃ + ỹ is also a nilpotent. Hence the result. Secondly, in a ring R̃, if
ũ is unit and x̃ is nilpotent, then ũ + x̃ is again a unit. If x̃ is nilpotent, then
x̃k = 0 for some k ∈ N. We have (ũ + x̃)(ũ−1 − ũ−2 x̃ + · · · + (−1)k−1 ũ−k x̃k−1 ) =
1 + (−1)k−1 ũ−k x̃k = 1, showing ũ + x̃ is a unit element in R̃. Hence the result.

Now suppose a0 is a unit element and a1 , a2 , · · · , an are nilpotent. Let f (x) =

a0 + a1 x + · · · + an xn . But ai are nilpotent implies ai xi are nilpotent polynomi-
als in R[x]. Also sum of two nilpotent elements is again a nilpotent, therefore
a1 x + a2 x2 + · · · + an xn is a nilpotent polynomial in R[x]. Let f (x) = a0 + g(x),
where g(x) = a1 x + · · · + an xn is a nilpotent polynomial in R[x]. But sum of
a unit element and a nilpotent is again a unit, therefore f (x) is a unit in R[x].
So f (x) has inverse in R[x]

Conversely, suppose f (x) = a0 + a1 x + · · · + an xn 6= 0 in R[x] has inverse,

therefore there exists g(x) = b0 + b1 x + · · · + bm xm 6= 0 in R[x] such that
f (x)g(x) = 1. Comparing constant terms of f (x)g(x) = 1, we have a0 b0 = 1.
Thus a0 is a unit element in R. Next we aim to show an is nilpotent. Comparing
coefficient of xn+m in the equation f (x)g(x) = 1, we have an bm = 0. We claim
aj+1
n bm−j = 0 for all 0 ≤ j ≤ m; and we establish our claim by induction over
j. For the base case j = 0, we need to show an bm = 0, which we have already
shown. Let aj+1 m bn−j = 0 for all 0 ≤ j ≤ i − 1. We aim to show that the
result is valid for j = i too. Comparing coefficient of xn+m−i in the equation
f (x)g(x) = 1, we have

an bm−i + an−1 bm−i+1 + · · · + an−i bm = 0

Multiplying by ain , and using induction hypothesis, we have ai+1 n bm−i = 0.

Thus result is valid for j = i too. So aj+1n bm−j = 0 for all 0 ≤ j ≤ m. For
j = m, we have am+1 n b0 = 0 ⇒ am+1
n = 0 as b0 a unit element in R. So an
is a nilpotent in R. Next we claim an−k is nilpotent for all 0 ≤ k ≤ n − 1.
We prove our claim by induction over k. When k = 0, we have already
seen an is a nilpotent as am+1n = 0. Suppose the result hold good for all
0 ≤ k ≤ i − 1, we will show the result holds good for k = i too. We have
f (x) unit element, and an−i+1 xn−i+1 , an−i+2 xn−i+2 , · · · , an xn as idempotent,
so f (x) − an−i+1 xn−i+1 − an−i+2 xn−i+2 − · · · − an xn is a unit element in R[x],
or a0 + a1 x + · · · + am−i am−i is a unit element. So proceeding as we did for
base case, we get (an−i )j+1 bm−j = 0 for all 0 ≤ j ≤ m. So for j = m, we have
am+1
n−i = 0 as b0 is a unit element. Thus we see result is valid for k = i too.
Thus an−k is nilpotent for all 0 ≤ k ≤ n − 1. In other words, a1 , a2 , · · · , an are
nilpotent elements. Thus f (x) = a0 + a1 x + · · · + an xn is a unit in R[x] if and
only if a0 is a unit and rest ai are nilpotent elements in R.

8. Prove that when F is a field, F [x1 , x2 ] is not a principal ideal ring.

Solution: We break the problem in two parts. First we will show for F being
a field implies F [x1 ] is not a field. Second, if F [x1 , x2 ] is a principal ideal ring
over F [x1 ], then F [x1 ] must be a field.

Suppose F [x1 ] is a field, with given that F is a field. Therefore if f (x1 ) 6=

0 ∈ F [x1 ], then f (x1 ) must be a unit element. But f (x1 ) = 1x1 ∈ F [x1 ] with
f (x1 )g(x1 ) 6= 1 ∀ g(x1 ) ∈ F [x1 ] as deg(f (x1 )g(x1 )) ≥ 1 while deg(1) = 0.
This means that f (x1 ) = x1 has no inverse which is not possible as F [x1 ] is
assumed to be a field. Hence F [x1 ] is not a field.

Next suppose F [x1 , x2 ] is a principal ideal ring over F [x1 ]. We define ψ :

F [x1 , x2 ] −→ F [x1 ] such that ψ(f0 (x1 ) + f1 (x1 )x2 + f2 (x1 )x22 + · · · ) = f0 (x1 ).
We left it to the reader to check φ so defined is well-defined and is an onto
ring-homomorphism. Thus F [x1 , x2 ]/Kψ ≈ F [x1 ], where Kψ is the kernel of
mapping ψ. Note that Kψ is an ideal in F [x1 , x2 ]. Also

Kψ = {f (x1 , x1 ) | ψ(f (x1 , x2 )) = 0}

= {0 + g1 (x1 )x2 + g2 (x1 )x22 + · · · | gi (x1 ) ∈ F [x1 ] ∀i}
= {x2 (g1 (x1 ) + g2 (x1 )x2 + · · · ) | gi (x1 ) ∈ F [x1 ] ∀i}
= {x2 g(x1 , x2 ) | g(x1 , x2 ) ∈ F [x1 , x2 ]}
= hx2 i

Now suppose, if possible there exist an ideal M ∈ F [x1 , x2 ] such that Kψ (

M ( F [x1 , x2 ]. But since F [x1 , x2 ] is assumed to be a principal ideal ring,
therefore M = h(x1 , x1 )F [x1 , x2 ] = hh(x1 , x1 )i as F [x1 , x2 ] is a commutative
ring with unity. Also since hx2 i ( hh(x1 , x2 )i, therefore x2 ∈ hh(x1 , x2 )i ⇒
x2 = h(x1 , x2 )q(x1 , x2 ) for some q(x1 , x2 ) ∈ F [x1 , x2 ]. Therefore

deg(x2 ) = deg(h(x1 , x2 )q(x1 , x2 ))

1 = deg(h(x1 , x2 )) + deg(q(x1 , x2 ))

So either deg(h(x1 , x2 )) = 1 and deg(q(x1 , x2 )) = 0, or deg(h(x1 , x2 )) = 0

and deg(q(x1 , x2 )) = 1. When deg(h(x1 , x2 )) = 1 and deg(q(x1 , x2 )) = 0,
x2 = h(x1 , x2 )q(x1 , x2 ) forces h(x1 , x2 ) = ux2 and q(x1 , x2 ) = u−1 , where u is a
unit element in R[x1 ]. But then M = hh(x1 , x2 )i = hux2 i = hx2 i = U which is
not the case. In the second case, with deg(h(x1 , x2 )) = 0 and deg(q(x1 , x2 )) = 1,
we have only possible solution as h(x1 , x2 ) = u and q(x1 , x2 ) = u−1 x2 . But
then M = hh(x1 , x2 )i = hui = h1i = F [x1 , x2 ] which contradicts our assump-
tion. Thus U is a maximal ideal of F [x1 , x2 ]. So F [x1 , x2 ]/U is a field. But
F [x1 , x2 ]/U ≈ F [x1 ]. So F [x1 ] is also a field.

Combing the two results, we have F [x1 , x2 ] is not a principal ideal ring for F
being a field.

9. Prove, completely, Lemma 3.11.2 and its corollary.

Solution: First we show the existence of greatest common divisor in a unique
factorization domain. Let R be some unique factorization domain and some
a, b ∈ R. We avoid a = b = 0 as greatest common divisor is not defined for it.
When either of them is 0 while other not, then trivially their greatest common di-
visor is the non-zero element itself. So we now assume both a and b are non-zero.
If any of them is unit, then their greatest common divisor is equal to 1 trivially.
So we also assume non of them is not unit too. But then R being a unique
β1 β2
factorization domain, so a = pα 1 α2 αn βn
1 p2 · · · pn and b = p1 p2 · · · pn , where αi ≥ 0
and βi ≥ 0, and pi are irreducible factors. Note that such representation with
common irreducible factors is possible as we have in this notation p0i = 1. We
define γi = Min(αi , βi ) ∀ i. Let d = pγ11 pγ22 · · · pγnn . Clearly d is uniquely(upto
associates) determined for the given a, b. We claim d = gcd(a, b). Clearly d | a
and d | b. Now suppose some c | a and c | b. But c | a implies c = pδ11 pδ22 · · · pδnn
with δi ≤ αi . But c | b also, so δi ≤ βi . But that means δi ≤ Min(αi , βi ). So
c | d. Also unique decomposition of a and b into irreducible pi guarantees their
greatest common divisor is uniquely determined upto the associates. Hence d is
the required greatest common divisor.

Next suppose gcd(a, b) = 1 and a | bc. We need to show a | c. If a is a unit, then

a | c trivially. So we assume a is not a unit. Clearly a6 | b as if a | b, then adding
a | a, we have a | gcd(a, b) = 1 which is not possible as a is not a unit. But if
a6 | b, then a | bc implies a | c as if a6 | c too, then a6 | bc too, which is not the case.
Hence a | c.

Finally, if a is irreducible and a | bc, then need to show a | b or a | c. In other

words, we need to show that in a unique factorization domain, every irreducible
element is a prime too. If either of b or c is a unit, then we have trivially a
dividing the non-unit element. So we assume a and b are non-unit elements. We
have a | bc implies bc = ad for some d ∈ R with d not a unit element otherwise
it would mean a is a reducible element which is not true. Also R being a unique
factorization domain, so b = p1 p2 · · · pn , c = q1 q2 · · · qm and d = r1 r2 · · · rk for
some irreducible pi , qi , ri . So we have

ar1 r2 · · · rk = p1 p2 · · · pn q1 q2 · · · qm = x(say)

x being an element in R must have a unique representation, therefore a = upi

for some unit u and some 1 ≤ i ≤ n, or a = u0 qj for some unit u0 and 1 ≤ j ≤ m.
When a = upi , we have pi = u−1 a ⇒ a | pi ⇒ a | b as pi | b for all i. So in this
case, a | b. When a = u0 qj , we have qj = u0−1 a ⇒ a | qj ⇒ a | c as qj | c for all j.
So in this case a | c. Thus we concluded a | bc implies a | b or a | c.

10. (a) If R is a unique factorization domain, prove that every f (x) ∈ R[x] an
be written as f (x) = af1 (x), where a ∈ R and wheref1 (x) is primitive.
(b) Prove that the decomposition in part(a) is unique (up to associates).
Solution:
(a) Let f (x) = a0 +a1 x+· · ·+an xn for some n ∈ W. Let a be the content of f (x).
Therefore a = gcd(a0 , a1 , · · · , an ). Note that existence of a, unique upto asso-
ciates is guaranteed as R is a unique factorization domain. Thus we have a | ai
for all i. So ai = aa0i for some a0i ∈ R. And so f (x) = aa00 +aa01 x+· · ·+aan xn =
a(a00 + a01 x + · · · + a0n xn ) = f1 (x)(say). Now suppose p be the content of f1 (x),
therefore p | a0i for all i. So a0i = pa00i for some a00i ∈ R. This leads to ai = apa00i .
So ap | ai ∀ i. But gcd(a0 , a1 , · · · , an ) = a, therefore ap | a ⇒ p | 1 ⇒ p is a
unit, which shows f1 (x) is primitive. Thus every f (x) ∈ R[x] can be written as
f (x) = af1 (x) where a is the content of f (x) and f1 (x) is primitive.

(b) Suppose decomposition is not unique(upto associates). Let f (x) = a1 f1 (x) =

a2 f2 (x). But then a1 , a1 , both are the greatest common divisors of non-zero
coefficients of f (x), so are associate of each other. Again note that the great-
est common divisor of elements exists uniquely(upto associates) in a unique
factorization domain. Thus a1 = ua2 for some unit u ∈ R. Using this,
a1 f1 (x) = a2 f2 (x) implies uf1 (x) = f2 (x). So a1 is associate of a2 and f1 (x) is
associate of f2 (x).

11. If R is an integral domain, and if F is its field of quotients, prove that any
element f (x) in F [x] can be written as f (x) = (f (x0 )/a), where f (x0 ) ∈ R[x]
and a ∈ R.
Solution: Let X ai
f (x) = xi
bi
i∈λ

where ai , bi ∈ R and λ is some finite index set such that abii 6= 0 ∀ i ∈ λ. Define
a = lcm({bi | i ∈ λ}). Therefore bi | a ∀ i ∈ λ ⇒ a = bi b0i for all i ∈ λ and
corresponding b0i ∈ R. So
X ai
f (x) = xi
bi
i∈λ
!
a X ai
= xi
a bi
i∈λ
!
1 X ai
= a xi
a bi
i∈λ
 !
1 X
= bi b0i xi
a
i∈λ
1 X
= f0 (x), where f0 (x) = bi b0i xi ∈ R[x]
a
i∈λ

Hence the result.

12. Prove the converse part of Lemma 3.11.4.

Solution: We are given f (x) ∈ R[x], and f (x) as an element of F [x] is irre-
ducible in F [x]. Suppose f (x) is reducible in R[x], therefore f (x) = g(x)h(x)
with neither g(x) nor h(x) is a unit or zero element in R[x]. But R[x] ⊂ F [x],
so g(x), h(x) ∈ F [x] too. Also if g(x), h(x) are not unit elements in R[x], then
they are also not unit elements in F [x]. But then the relation f (x) = g(x)h(x)
implies f (x) is reducible in F [x] which is not true. Hence f (x) is irreducible in
R[x] too.

13. Prove corollary 2 to Theorem 3.11.1.

Solution: If F is a field then F [x1 ] is a Euclidean domain (Theorem 3.9.1).
But a Euclidean domain is also a unique factorization domain (Theorem 3.7.2).
So F [x1 ] is a unique factorization domain. But F [x1 ] being a unique factor-
ization domain implies F [x1 , x2 ] is also a unique factorization domain. Again
F [x1 , x2 ] being a unique factorization domain implies F [x1 , x2 , x3 ] is a unique
factorization domain. Continuing this way, we get F [x1 , x2 , . . . , xn ] is a unique
factorization domain.

14. Prove that a principal ideal ring is a unique factorization domain.

Solution: First we will prove two lemmas required to prove the main result.

Lemma 1: In a principal ideal ring, an element is prime if and only if it is

irreducible. Let R be some principal ideal ring. First suppose a is prime. So
a is neither a unit or zero element. Suppose a = bc, therefore a | (bc). But a
being prime implies a | b or a | c. When a | b, it implies b = ad for some d ∈ R.
Therefore a = bc = (ad)c = a(dc) ⇒ a(1 − dc) = 0 ⇒ 1 − dc = 0 ⇒ dc = 1.
So a | b implies c is a unit. Similarly a | c implies b is a unit. So if a = bc
implies either b or c is unit. Thus we concluded a is an irreducible element.
Conversely suppose a is an irreducible in R. Suppose a | (bc). We define
U = {ax + by | x, y ∈ R}. Easy to check that U is an ideal of R. But R
being a principal ideal ring, so U = dR for some d ∈ R. But a ∈ U , so a = dr
for some r ∈ R. Now a being an irreducible element so a = dr implies ei-
ther d is a unit or r is a unit. When d is a unit, we have U = dR = R. So
1 ∈ U . Therefore we have 1 = ar1 + br2 for some r1 , r2 ∈ R. But that means
c = c(ar1 ) + c(br2 ) ⇒ c = a(cr1 ) + (bc)r2 ⇒ c = a(cr1 + r2 ) ⇒ a | c. So at least
a | c (a may divides b too). On the other hand when r is a unit, U = dR = aR.
But since b ∈ U , therefore b = ar3 for some r3 ∈ R. So at least a | b in this case.
Thus a | bc implies a | b or a | c. So if a is an irreducible then it is a prime too.
Hence the lemma.

We call a sequence of ideals U1 , U2 , · · · as strictly increasing sequence if Un (

Un+1 ∀n ∈ N. We claim in a principal ideal ring, any strictlySincreasing se-
quence of ideals U1 , U2 , · · · must be finite in length. Let U = Ui . We first
i
claim U is an ideal. Let some x, y ∈ U , therefore x ∈ Um and y ∈ Un for some
ideals Um , Un in the strictly increasing sequence of ideals. So either Um ( Un
or Un ( Um . In first case, when Um ( Un , we have x − y ∈ Un ⊂ U . Also in
second case when Un ( Um , we have x−y ∈ Um ⊂ U . So x−y ∈ U ∀ x, y ∈ U .
Next, let some x ∈ U , therefore x ∈ Um for some ideal in the strictly increasing
sequence. But xr ∈ Um ⊂ U ∀ r ∈ R. Thus we have shown U is an ideal of
R. But R being a principal ideal ring, therefore U = aR for some a ∈ R. But
1 ∈ R, therefore a1 ∈ U . So a ∈ Uk for some ideal Uk in the strictly increasing
sequence of ideals. But then U = aR ⊂ Uk , implying that Uk is the last member
of the strictly increasing sequence of ideals. Hence in a principal ideal ring, any
strictly increasing sequence of ideals, U1 , U2 , · · · must be finite in length.

Lemma 2: In a principal ideal ring, any strictly increasing sequence of ideals,

U1 , U2 , · · · must be finite in length.

Now with the above lemmas at our disposal, we aim to prove that every prin-
cipal ideal ring is a unique factorization domain. Suppose R be some principal
ideal ring. Let some a ∈ R with not a unit or zero-element. First we will
show a is a product of irreducible elements. To prove so, we first show a has
at least one irreducible factor. Now if a itself is an irreducible element, then
we are done. But if not, then a = b1 a1 , where neither b1 or a1 is a unit or
zero-element. Now again if a1 is an irreducible element then we are done, but
if a1 is not an irreducible element, then a1 = b2 a2 , where neither b2 or a2 is
a unit or zero-element. Continuing this way we get a sequence a1 , a2 , · · · with
each element neither a unit nor a zero-element. From this sequence we induce
a sequence of ideals, a1 R, a2 R, · · · . We claim this sequence of ideals is strictly
increasing, i.e. a1 R ( a2 R ( · · · . Clearly an R ⊂ an+1 R as an = bn+1 an+1 .
But an R = an+1 R implies an+1 = an r for some r ∈ R. But that means
an+1 = bn+1 an+1 r ⇒ an+1 (1 − bn+1 r) = 0 ⇒ bn+1 r = 1 as R is an integral
domain. So bn+1 is a unit, a contradiction, implying an R ( an+1 R. So we have
a strictly increasing sequence of ideals, a1 R, a2 R, · · · . But Lemma 2 implies this
sequence of ideals must terminate for some ak . But that means ak cannot be
factorized into product into bk+1 ak+1 with both bk+1 and ak+1 not equal to 0
or unit. But that means ak is an irreducible element. So we have shown that
a must have an irreducible factor. We now show a is a product of irreducibles.
We have a = p1 c1 where p1 is an irreducible element as we have just shown a
must have at least one irreducible factor. Clearly c1 6= 0, otherwise a = 0, which
is not the case. Also if c1 is a unit, then we are done, but if not, then c1 = p2 c2 ,
where p2 is an irreducible element as c1 too must have at least one irreducible
factor. Continuing this way, we get a sequence c1 , c2 , · · · . From this sequence we
induce a sequence of ideals as c1 R, c2 R, · · · . We claim this sequence of ideals is
strictly increasing, i.e. cn R ( cn+1 R. To establish our claim, we first note that
(pn+1 cn+1 )R ⊂ cn+1 R. Also if (pn+1 cn+1 )R = cn+1 R ⇒ cn+1 = pn+1 cn+1 r
for some r ∈ R, or cn+1 (1 − pn+1 r) = 0. R being an integral domain im-
plies 1 = pn+1 r. But that means pn+1 is a unit, which is not the case. So
cn R ( cn+1 R. Thus the sequence of ideals c1 R, c2 R, · · · is strictly increas-
ing. But Lemma 2 implies it must terminate for some cj . But that means
cj cannot be factorized into the product pj+1 cj+1 as otherwise we would have
cj R ( cj+1 R. But that also means cj is an irreducible element. Thus we have
a = p1 p2 · · · pj cj , with all factors as irreducibles.

What left to be shown is the uniqueness of this factorization upto the associates
and the order in which the irreducible factors appear. Suppose a = p1 p2 · · · pm =
q1 q2 · · · qn . We use induction over m, i.e number of irreducible factors, to show
the uniqueness. When m = 1, we have a = p1 , i.e. a is an irreducible. Therefore,
a = q1 q2 · · · qn implies n = 1 and a = q1 . Therefore for m = 1, we have n = 1
and p1 = q1 . Suppose the uniqueness(upto the associates and the order) holds
good for m = i − 1. We need to show result is valid for m = i too. We have
a = p1 p2 · · · pi = q1 q2 · · · qn . But since p1 | a, therefore p1 | (q1 q2 · · · qn ). But
since p1 is irreducible, so by Lemma 1, p1 is prime too. Therefore p1 divides
some qi , say q1 . But q1 itself is irreducible, therefore q1 = p1 u for some unit
u. Now we have up1 p2 · · · pi = uq1 q2 · · · qn = q1 (uq2 · · · qn . R being an integral
domain, so
p2 p3 · · · pi = uq2 q3 · · · qn
But the induction hypothesis implies factors on both sides are same upto the
associates and the order in which they appear as number of irreducible factors
are i − 1. Also up1 = q1 . So a is unique upto the associates and the order in
which its factors appear for m = i too. Hence a is uniquely decomposable into
irreducible factors in general. Thus we concluded R is a unique factorization
domain.

15. If J is the ring of integers, prove that J[x1 , . . . , xn ] is a unique factorization

domain.
Solution: J, ring of integers, is a unique factorization domain, so using Corol-
lary 2 of theorem 3.11.1, we have J[x1 , x2 , . . . , xn ] is also a unique factorization
domain.