Homepage Solution manuals David S. Dummit Abstract Algebra Exercise 4.3.28(Groups of order $pq$ as permutation groups)

Exercise 4.3.28(Groups of order $pq$ as permutation groups)

Let p and q be primes with p < q . Prove that a non-abelian group G of order 𝑝𝑞 has a non-normal subgroup of index q , so that there exists an injective homomorphism into S q . Deduce that G is isomorphic to a subgroup of the normalizer in S q of the cyclic group generated by the q -cycle ( 1 2 q ) .

Answers

Proof. Let G be a non-abelian group G of order n = 𝑝𝑞 (where p < q and p , q are prime numbers).

By Cauchy’s Theorem, there exists an element a of order q in G , so H = a is a subgroup of order q , and | G : H | = p . Since p is the smallest prime dividing n = | G | , then H G by Corollary 5 (p.120).

Moreover there is an element b of order p in G , so K = b is a subgroup of order p and index q .

Assume for the sake of contradiction that K is normal in G .

By Lagrange’s Theorem | H K | divides p and q , so | H K | = 1 and H K = { 1 } . Then by Proposition 13 p. 93, | 𝐻𝐾 | = | H | | K | | H K | = 𝑝𝑞 = | G | , so G = 𝐻𝐾 . In summary

(i)
G = 𝐻𝐾 ,
(ii)
H K = { 1 } ,
(iii)
H G , K G .

Consider the map

u { H × K G ( h , k ) h𝑘 .

By (i), u is surjective. If u ( h , k ) = u ( h , k ) , then h𝑘 = h k , thus h 1 h = k k 1 H K = { 1 } by (ii), therefore ( h , k ) = ( h , k ) , so u is injective.

Moreover, if h H and k K , then by (iii), h𝑘 h 1 k 1 = ( h𝑘 h 1 ) k 1 = h ( k h 1 k 1 ) H K = { 1 } , thus h𝑘 = 𝑘h . If ( h , k ) H × K and ( h , k ) H × K , then

u ( h , k ) u ( h , k ) = h𝑘 h k = h h k k = u ( h h , k k ) = u ( ( h , k ) ( h , k ) ) ,

so u is an isomorphism, and G H × K . Since H and K are cyclic, G is abelian, in contradiction with the hypothesis. This contradiction shows that K is not normal in G .

Now consider the action of G on the set A of left cosets of K , where | A | = | G : K | = q . This action affords a representation π K : G S A , defined by

π K ( g ) ( 𝑥𝐾 ) = 𝑔𝑥𝐾 ( g , x G ) . (1)

Recall that H = a , where | a | = q . We claim that

A = { K , 𝑎𝐾 , a 2 K , , a q 1 K } . (2)

Indeed these left cosets of K are distinct: if a i K = a j K , where 0 i j < q , then a j i K H = { 1 } , therefore a j i = 1 , where 0 j i < q , so i = j . Since there are q = | A | such cosets, this proves (2).

Consider the bijection f : [ [ 1 , q ] ] A given by

k 1 2 q
f ( k ) 𝑎𝐾 a 2 K a q K = K

i.e.,

f ( k ) = a k K ( 1 k q ) . (3)

This labelling of the cosets affords a permutation representation φ : G S q , defined by

j = φ ( g ) ( i ) f ( j ) = π K ( g ) ( f ( i ) ) ( i , j [ [ 1 , q ] ] ) . (4)

Then f φ ( g ) = π K ( g ) f for all g G , so that the following diagram is commutative

Writing φ g = φ ( g ) S q , the definitions (2) and (3) imply

j = φ g ( i ) a j K = g a i K ( i , j [ [ 1 , q ] ] ) . (5)

Then

φ ( g ) φ ( g ) = f 1 π K ( g ) f f 1 π K ( g ) f = f 1 π K ( g ) π K ( g ) f = f 1 π K ( g g ) f = φ ( g g ) ,

so φ : G S q is a homomorphism.

By Theorem 3 p. 119,

ker ( π K ) = x G 𝑥𝐾 x 1 .

Since K is not normal in G , there is some y G such that 𝑦𝐾 y 1 K . Therefore L = K 𝑦𝐾 y 1 is a subgroup of K , and L K , otherwise K 𝑦𝐾 y 1 , where | K | = | 𝑦𝐾 y 1 | , so 𝑦𝐾 y 1 = K , which contradicts the definition of y . Since the order of K is prime, this shows that K 𝑦𝐾 y 1 = { 1 } for some y G . Since ker ( π K ) K 𝑦𝐾 y 1 = { 1 } , we obtain

ker ( π K ) = x G 𝑥𝐾 x 1 = { 1 } .

So the action of G on A is faithful, and π K : G S A is an injective homomorphism.

If φ ( g ) = ( ) = id [ [ 1 , q ] ] , then π K ( g ) = f φ ( g ) f 1 = id A , thus g = 1 . This shows that φ is an injective homomorphism.

Put 𝜃 = ( 1 2 3 q ) . We compute φ a and φ b .

By (5), for all i , j [ [ 1 , q ] ] ,

j = φ a ( i ) a j H = a a i H a j H = a i + 1 H j i + 1 ( 𝑚𝑜𝑑 q )

Therefore

φ a = ( 1 2 3 q ) = 𝜃 .

(So 𝜃 corresponds to the element a G in the permutation representation.)

Since φ a = 𝜃 𝜃 , φ a 𝜃 N S n ( 𝜃 ) , so

φ a N S n ( 𝜃 ) . (6)

To compute φ b , note that 𝑏𝑎 b 1 H , since H = a G . Therefore there is some fixed integer k such that

𝑏𝑎 b 1 = a k , where  0 k < q .

Moreover, by (2), there is some fixed integer l so that b a l K , thus

𝑏𝐾 = a l K .

Then for all integers i , b a i b 1 = ( 𝑏𝑎 b 1 ) i = ( a k ) i = a 𝑘𝑖 . Hence for all i , j [ [ 1 , q ] ]

j = φ b ( i ) a j K = b a i K a j K = a 𝑘𝑖 𝑏𝐾 a j K = a 𝑘𝑖 + l K j 𝑘𝑖 + l ( 𝑚𝑜𝑑 q ) .

This shows that

φ b = ( 1 2 q [ k + l ] [ k 2 + l ] [ 𝑘𝑞 + l ] ) = ( i [ 𝑘𝑖 + l ] )

where [ j ] denotes the representative of the class of j modulo q such that 1 [ j ] q . (Note : the permutation group of such permutations is known as AGL ( 1 , 𝔽 q ) S q . This group was studied by Lagrange and Galois.)

It remains to prove that φ b N s q ( 𝜃 ) . Using the notation [ j ] , we obtain for all i [ [ 1 , q ] ] ,

( φ b 𝜃 ) ( i ) = φ b ( [ i + 1 ] ) = [ k ( i + 1 ) + l ] = [ φ b ( i ) + k ] = 𝜃 k ( φ b ( i ) ) .

So φ b 𝜃 = 𝜃 k φ b , and

φ b 𝜃 φ b 1 = 𝜃 k .

More generally, for all integers i , φ b 𝜃 i φ b 1 = 𝜃 𝑖𝑘 𝜃 . This shows that φ b normalizes 𝜃 , so

φ b N S q ( 𝜃 ) . (7)

Let G ~ = φ ( G ) S q . Since φ is an injective homomorphism, G ~ is a permutation group isomorphic to G , and

G ~ = φ ( G ) = φ ( a , b ) = φ ( a ) , φ ( b ) = φ a , φ b .

By (6) and (7), φ a , φ b N S q ( 𝜃 ) , so

G G ~ N S q ( 𝜃 ) :

G is isomorphic to a subgroup of the normalizer in S q of the cyclic group generated by the q -cycle ( 1 2 q ) . □

Note 1: The normalizer in S q of the cyclic group generated by the q -cycle 𝜃 = ( 1 2 q ) is the whole group AGL ( 1 , 𝔽 q ) (see David Cox, “Galois Theory" Lemma 14.1.2 p. 414).

Note 2. Cox proves a generalization of the preceding result in Proposition 14.1.4:

“Every solvable transitive subgroup G S p is conjugate to a subgroup of AGL ( 1 , 𝔽 p ) containing 𝜃 .”

This is part of the proof of Galois Theorem on solvable irreducible polynomials of prime degree p (Cox, Theorem 14.1.1 p. 413)

User profile picture
2026-02-13 11:10
Comments