Math 504 HW 4

02 NOV 2021

Generators of Sn

Proposition 2.4 The symmetric group Sn can be generated by two elements: a cycle of length 2 and
a cycle of length n.
Proof. Let σ = (12) and let τ = (12 · · · n). We need to show that any element can be expressed as a finite
combination of σ, τ, σ −1 , and τ −1 . Since σ −1 = σ then we only need to show that for all ϕ ∈ Sn

ϕ = τ k σ(τ −1 )j

for 0 ≤ k, j ≤ n. First we show that ϕ can be written as a product of disjoint cycles. We impose the fol-
lowing algorithm on ϕ. Let α1 be the first integer such that ϕ(α1 ) ̸= α1 and write (α1 . Then let α2 = ϕ(α1 )
and compute ϕ(α2 ).

If ϕ(α2 ) = α1 then we have our first cycle, namely (α1 α2 ). Now we move on to the next element
̸ α1 , α2 such that ϕ(β1 ) ̸= β1 and repeat the process we did for α1 but now with β1 .
β1 =

If ϕ(α2 ) = α3 ̸= α1 then write (α1 α2 and start again by looking at whether ϕ(α3 ) = α1 or if
ϕ(α3 ) = α4 ̸= α1

Since we are dealing with bijections on {1, 2, . . . , n} then the process has to end with either generating
a cycle or a product of disjoint cycles. As an example consider S5 and consider the bijection:

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

Then using our algorithm we get the first cyce as (152) then we start again with 3 and get our last
cycle (34) and conclude that ϕ = (34)(152)

If ψ = (a1 a2 . . . ak ) is a cycle then we can decompose it as ψ = (a1 a2 )(a2 a3 ) · · · (ak−1 ak ). Now we

consider the identity that
τ k (12)τ −k = (τ k (1)τ k (2))
All that τ does is shift elements 1, 2, . . . n to its successor so if we apply this k-times then all we are doing
is shifting an element k times around the cycle. With this in mind we see that (τ k (1)τ k (2)) = (1 + k, 2 + k)
and so by varying k from 1 to n we can generate all 2-cycles of the form (i, i + 1). All that is left to show

1
is that any transposition can be generated by transpositions of the form (i, i + 1). So we consider (x, y)
such that y − x ̸= 1. We write
(x, y) = (x, x + 1)(x + 1, y)(x, x + 1)
Now we look at the middle term and if y − (x + 1) ̸= 1 then we just rewrite it as (x + 1, x + 2)(x + 2, y)(x +
1, x + 2). Clearly the process will end since y − x is some finite number less than n. So we see that any
2-cycle can be decomposed into a product of 2-cycles of the form (i, i + 1).

Altogether we have that any 2-cycle of the form (i, i + 1) can be written in terms of τ k (12)τ −k , all
2-cycles can be generated by 2-cycles of the form (i, i + 1), all cycles can be generated by 2-cyclces, and
any permutation can be generated by disjoint cycles. Hence, any permutation can be expressed in terms
of σ = (12) and τ = (12 · · · n).
Proposition 2.5 Any permutation σ ∈ Sn can be written as a composition of disjoint cycles.
Proof. The result was shown in the proof of Proposition 2.4
Theorem 2.8 Let σ, τ ∈ Sn . Then σ and τ are conjugate if and only if their decompositions into
disjoint cycles can be put into one-to-one correspondence such that the correspodning cycles are the same
length.
Proof. Suppose there exists some ϕ ∈ Sn such that ϕσϕ−1 = τ . Let ϕ = ϕ1 · · · ϕa , τ = τ1 · · · τs , and
σ = σ1 · · · σr be the cycle decompostions of σ, ϕ and τ . Consider just a single cycle σr = (a1 · · · ak ) and
lets look at the first composition in ϕσϕ−1 which will be
σr (ϕ−1 )
We’ll consider an element ai in the cycle of σr . Let b = ϕ(ai ) and c = ϕ(a(imodk)+1 ). We then have
σr (ϕ−1 (b)) = σr (ai ) = a(imodk)+1

Since all the cycles in the decomposition for σ are disjoint then
(σ1 · · · σr−1 )(a(imodk)+1 ) = a(imodk)+1
This implies that (ϕσϕ−1 )(b) = ϕ(a(imodk)+1 ) = c. With this in mind we can just write
ϕσϕ−1 = (ϕσ1 ϕ−1 ) · · · (ϕσr ϕ−1 )
We know that each of these terms can be written as
ψi = ϕσi ϕ−1 = (ϕ(a1j ) · · · ϕ(akj ))
where j = 1, . . . r and kj is the length of the cycle for σj .

This forms a disjoint set of cycles since ϕ is a bijection. Now we can equate
τ1 · · · τs = ψ1 · · · ψr
We have two products of disjoint cycles that must be equal, which can only be true if r = s and so σ and
τ must have the same cycle decompostion length, and for each τi there must correspond a σj such that
ϕσj ϕ−1 = τi which means σj and τi have the same cycle length, furthermore
τi = (ϕσj ϕ−1 )

2
 Now suppose we have a one-to-one correspondence between the cycle decomposition of σ and that for
τ then consider some integer x ∈ {1, . . . , n}. Since σ is a bijection then x must belong to some cycle σi
and since there corresponds a cycle τi then there corresponds an integer y in the cycle for τi , in the same
place. Define a permutation ϕ such that ϕ(x) = y. Letting σi = (a1 · · · x · · · ak ) then there corresponds
τi = (b1 · · · y · · · bk ). We just do this for every element in {1, . . . , n}, for each σi ϕ just sends ai to it’s
corresponding value in τi

We see that ϕσi ϕ−1 = τi and furthermore

ϕσϕ−1 = (ϕσ1 ϕ−1 ) · · · (ϕσr ϕ−1 ) = τ1 · · · τr = τ

Proposition 2.11 The Symmetric group Sn is generated by transpositions.

Proof. It will suffice to show that all cycles can be generated by transpositions, which follows easily since
if σ = (a1 a2 · · · ak ) is a cycle then we can write

σ = (a1 a2 )(a2 a3 ) · · · (ak−1 ak )

Since every permutation can be written as a product of disjoint cycles then it follows that any permutation
can be written as a product of transpositions.
Alternating Group

Proposition 3.1 The map Sgn : Sn 7→ Z/2Z, Sgn(σ) = −1 if σf = −f and Sgn(σ) = 1 of σf = f ,

where Y
f (x1 , . . . xn ) = (xi − xj )
for i < j, is a group homomorphism.
Proof. Consider the group {−1, 1}, with the operation of multiplication, which is isomporphic to Z/2Z.
We’ll show that Sgn : Sn 7→ {−1, 1} is a group homomorphism.

If σ1 f = f and σ2 f = f then σ1 (σ2 f ) = σ1 f = f =⇒ Sgn(σ1 σ2 ) = 1 = Sgn(σ1 )Sgn(σ2 ).

If σ1 f = −f and σ2 f = −f then σ1 (σ2 f ) = σ1 (−f ) = f =⇒ Sgn(σ1 σ2 ) = 1 = Sgn(σ1 )Sgn(σ2 )

Suppose that σ1 f = f and σ2 f = −f then σ1 (σ2 f ) = σ1 (−f ) = −f =⇒ Sgn(σ1 σ2 ) = −1 =

Sgn(σ1 )Sgn(σ2 )

So we see that in any case Sgn : Sn 7→ {−1, 1} is a group homomorphism.

Corollary 3.3 The subset of all even permutations is a normal subgroup of Sn
Proof. Since {1} is normal in {−1, 1} then Sgn−1 ({1}) = An is normal in Sn since Sgn is a group
homomorphism.
Theorem 3.5
(1) If τ ∈ Sn is a transposition, then Sgn(τ ) = −1
(2) A permutation is σ is even if and only if it can be written as a product of an even number of transpo-
sitions.

3
Proof. (1) Let τ = (ab) with a < b. We can just factor out the term xa − xb from f to get
Y
f (x1 , . . . xn ) = (xa − xb ) (xi − xj )
Then we have Y Y
τ f = (xb − xa ) (xi − xj ) = −(xa − xb ) (xi − xj ) = −f

So we have that Sgn(τ ) = −1

(2) Suppose that σ is even, then Sgn(σ) = 1. Since σ can be written as a finite product of transpositions
and Sgn is a group homomorphism then
Sgn(σ) = Sgn(σ1 · · · σk ) = Sgn(σ1 ) · · · Sgn(σk ) = (−1)k
Since Sgn(σ) = 1 then k must be even.

Conversely suppose that σ = σ1 · · · σk is a product of an even number of transpositions. Then

Sgn(σ) = Sgn(σ1 · · · σk ) = Sgn(σ1 ) · · · Sgn(σk ) = (−1)k = 1
since k is even. Hence, σ is even.
Theorem 3.6 The group An is generated by 3-cycles of the form (12i), 3 ≤ i ≤ n
Proof. Let σ ∈ An then σ = σ1 · · · σk where k is even. Consider σi = (ai bi ) and σi+1 = (ai+1 bi+1 ). If
ai = ai+1 then
(ai bi )(ai bi+1 ) = (ai bi+1 bi )

If the two transpostions are disjoint then

(ai bi )(ai+1 bi+1 ) = (ai bi )(bi bi+1 )(bi bi+1 )(ai+1 bi+1 )
= (ai bi bi+1 )(bi bi+1 ai+1 )
So any two pairs of transpositions can be written as a 3 cycle or the product of 2 3-cycles. Let ψ be a
3-cycle in Sn . Consider the following

If ψ contains 1 and 2 then it can be written as either (12i) or (12i)2 = (1i2)

If ψ contains 1 but not 2 then let ψ = (1bc), we have

(1bc) = (1c)(1b) = (1c)(12)(12)(1b) = (12c)(12b)
The case for if ψ contains 2 but not 1 is identical.

If ψ contains neither 1 or 2 then let ψ = (abc), we have

(abc) = (ab)(bc)
= (ab)(1b)(1b)(bc)
= (1ab)(1bc)
= (12b)(12a)(12c)(12b)
So we see that any 3-cycle can be generated by 3-cycles of the form (12i) and since the product of pairs
of transpositions can be written in terms of 3-cycles , we conclude that all of An is generated by 3-cycles
of the form (12i) where 3 ≤ i ≤ n.

4
 Derived Series for Sn

Theorem 4.1 The symmetric group Sn is solvable for n = 1, 2, 3, 4.

Proof. For n = 1 we have S1 = {(1)}. So G0 = S1 = e.

For n = 2 we have S2 = {(1), (12)}. G0 = S2 , and G1 = [S2 , S2 ]. Since

(12)−1 (1)−1 (12)(1) = (12)(1)(12)(1) = (12)(12)(1) = (1)

then we see that [S2 , S2 ] = {(1)} = e. So =S2 is solvable with derived series

e = G1 , G0 = S2

For n = 3 we have S3 = {(1), (12), (13), (23), (123), (132)}. [S3 , S3 ] = {(123), (132), (1)} since

(12)(13)(12)(13) = (132)(213) = (132)2 = (123)

and similarly
(13)(12)(13)(12) = (123)2 = (132)
Since elements in S3 are either 2-cycles or 3-cycles then

x−1 y −1 xy

will always be even, hence [S3 , S3 ] cannot have any transpositions so it must only contain the 3-cycles,
which are precisely the elements of A3 . So we see that [S3 , S3 ] = A3 . Now we consider [A3 , A3 ]. Since
(123)−1 = (132) then the only element in the commutator is (1) = e. Hence, [A3 , A3 ] = e. So, S3 is solvable
with derived series
G0 = S3 , G1 = A3 , G2 = e

For n = 4 we see that x−1 y −1 xy will always be even because xy will always contain the same number
of transpositions as x−1 y −1 so then x−1 y −1 xy will always have twice that many transpositions which is
an even number. So we conclude that [S4 , S4 ] = A4 . Next we consider [A4 , A4 ]. Since the inverse of any
3-cycle is just another 3-cycle then no matter what x or y is we wont get out another 3-cycle. The only
possible elements that can be in the commutator are the products of disjoint 2-cycles, hence

[A4 , A4 ] = {(1), (12)(34), (13)(24), (14)(23)}

Since each element is self inverse and has order 2 we see that

x−1 y −1 xy = xyxy = (xy)2 = (1)

So we have that [[A4 , A4 ], [A4 , A4 ]] = e and so S4 is solvable with derived series

G0 = S4 , G1 = A4 , G2 = [A4 , A4 ], G3 = e

5
 Theorem 4.2
(1) [Sn , Sn ] = An
(2) For n ≥ 5 [An , An ] = An

Proof. (1) By definiton [Sn , Sn ] =< x−1 y −1 xy > for x, y ∈ Sn . We write x and y as products of 2 cycles
x = x1 · · · xk and y = y1 , · · · yj .
x−1 = (x1 · · · xk )−1 = (xk · · · x1 )
and similarly
y −1 = (y1 , · · · yj )−1 = yj · · · y1
Then we have x−1 y −1 xy = (xk · · · x1 )(yj · · · y1 )(x1 · · · xk )(y1 , · · · yj ). This is an even permutation since it
is the prooduct of 2(k+j) transpositions. So we see that the commutator subgroup of Sn is contained in An .

Now, suppose that x and y belong to An then x = x1 · · · xk and y = y1 , · · · yj where k and j are both
even numbers. Then x−1 y −1 xy will always give us an element of An . So if we let x and y run through all
elements of An then we have to get back every element of An . So we see that every element of An is a
commutator and so An is contained in [Sn , Sn ]. We conclude that An = [Sn , Sn ].

(2) [An , An ] =< x−1 y −1 xy > for x, y ∈ An . Clearly [An , An ] is contained in An .

Conversely suppose that x ∈ An . As shown earlier any element of An can be generated by 3-cyclces of
the form (12i) for i ≥ 3. Consider then a 3-cycle x = (12i). Let y = (1i2) and z = (i2)(i + 1, i + 2). Then
we have
yzy −1 z −1 = y(z(12i)z −1 ) = y(z(1), z(2), z(i)) = y(1i2) = (12i)
So every 3-cycle of the form (12i) is in the commutator. Since any element of An can be written in
terms of such 3-cycles then every element of An is a commutator and so An is contained in the commutator
subgroup. We conclude An = [An , An ]
Theorem 4.4 For n ≥ 5, An is simple.
Proof. First we claim that if N is a normal subgroup of An that contains a 3-cycle then N = An . Since
any 3-cycle can be written in terms of 3-cycles of the form (12i) for i ≥ 3, then we assume that N contains
an element of the form (12i). Consider τ = (12)(ik). We have
τ (12i)2 τ −1 = (τ (1), τ (i), τ (2)) = (2k1) = (12k)
Since N is normal then τ (12i)2 τ −1 ∈ N . If we just let k range from 3 to n then we essentially produce
all 3 cycles of the form (12k) and hence all of An . We conclude that N = An .

Now let x be an element of a non-trivial normal subgroup N in An . There are five possible structures of x

Case 1: If x is a 3 cycle then as just shown, N = An

Case 2: If x is the product of disjoint cycles where at least one is of length greater than 3 then we write
x = y(a1 · · · ak ) where y is just the product of the other disjoint cycles. If we consider the element in N
x−1 (a1 a2 a3 )x(a1 a2 a3 )−1 = [(a1 · · · ak )y]−1 (a1 a2 a3 )y(a1 · · · ak )(a1 a3 a2 )
= (ak · · · a1 )y −1 y(a1 a2 a3 )(a1 · · · ak )(a1 a3 a2 )
= (ak · · · a1 )(a1 a2 a3 )(a1 · · · ak )(a1 a3 a2 )
= (ak a1 a2 )(a1 a3 a2 )
= (a1 a3 ak )

6
 So N containts a 3-cycle and thus N = An .

Case 3: If x is a disjoint product of 3-cycles then we write x = y(a1 a2 a3 )(a4 a5 a6 ) where y is just the
product of the rest of the disjoint 3-cycles. Consider the element in N

x−1 (a1 a2 a4 )x(a1 a2 a4 )−1 = [(a4 a6 a5 )(a1 a3 a2 )y −1 ]y(a1 a2 a4 )(a1 a2 a3 )(a4 a5 a6 )(a1 a4 a2 )
= (a4 a6 a5 )[(a1 a3 a2 )(a1 a2 a4 )(a1 a2 a3 )](a4 a5 a6 )(a1 a4 a2 )
= [(a4 a6 a5 )(a3 a1 a4 )(a4 a5 a6 )](a1 a4 a2 )
= (a3 a1 a6 )(a1 a4 a2 )
= (a1 a4 a2 a6 a3 )

So N contains a cycle of length greater than 3, hence by case 1 N = An

Case 4: If x is the disjoint product of a 3-cycle and disjoint 2-cycles then we write x = y(a1 a2 a3 ) where
y is just the disjoint product of the 2 cycles. Then

x2 = (y(a1 a2 a3 ))(y(a1 a2 a3 )) = y 2 (a1 a2 a3 )2 = (a1 a3 a2 )

So we conclude that N = An since.

Case 5: If x is the product of disjoint 2 cycles then write x = y(a1 a2 )(a3 a4 ) where y is just the product
of the rest of the disjoint 2-cycles. Consider the element of N

x−1 (a1 a2 a3 )x(a1 a3 a2 ) = (a3 a4 )(a1 a2 )y −1 y(a1 a2 a3 )(a1 a2 )(a3 a4 )(a1 a3 a2 )

= (a3 a4 )[(a1 a2 )(a1 a2 a3 )(a1 a2 )](a3 a4 )(a1 a3 a2 )
= [(a3 a4 )(a2 a1 a3 )(a3 a4 )](a1 a3 a2 )
= (a2 a1 a4 )(a1 a3 a2 )
= (a2 a1 )(a1 a4 )(a2 a1 )(a1 a3 )
= (a2 a4 )(a1 a3 )

Since n ≥ 5 then let z = (a1 a3 b) where b ̸= a1 , a2 , a3 , a4 . Consider the element of N

z −1 (a2 a4 )(a1 a3 )z(a2 a4 )(a1 a3 ) = (a1 ba3 )[(a2 a4 )(a1 a3 )(a2 a4 )](a1 a3 b)(a1 a3 )
= (a1 ba3 )[(a1 a3 )(a1 a3 b)(a1 a3 )]
= (a1 ba3 )(a3 a1 b)
= (a1 a3 b)

So N contains a 3-cycle and N = An . We conclude that if n ≥ 5 then An is simple.

7
 Sylow Subgroups of Spn

Lemma 5.1 The maximal power or p that divides pn !, denoted v(n), is equal to 1 + p + . . . + pn−1 .
Proof. We write pn ! = 1 × 2 × · · · × pn

For r = 1, . . . n − 1 consider pr which divides pn We know that we can also write pn ! = pa m where m
is the product of all ther terms berrween 1 and pn that are not multiples of p. Counting exponents from
each power of p we get
pn pn pn pn
a= + 2 · · · + n−1 + n = pn−1 + · · · + p + 1
p p p p
pn n n n
So the maximal power of p that divides pn ! is v(n) = a = p
+ pp2 · · · + pn−1
p
+ ppn = pn−1 + · · · + p + 1
Proposition 5.2 The symmetric group Spn has a sylow p-subgroup.
Proof. Let G = Spn . Consider the set S denoting all the subsets of G of cardinality pk where |G| = pk m
k

and p ∤ m. The order of S is then |S| = ppkm . Define the action of G on S by left multiplication

(a, T ) 7→ aT

where a is in G and T is in S.

Consider the class equation on S

|S| = |SG | + |OT1 | + · · · + |OTk |

where SG is the set of fixed points of S and OTi is the nontrivial equivalence class or orbit for represen-
tative Ti . Since p ∤ |S| and each OTi must divide |S| then there must exist at least one OTi such that p ∤ |OTi |.

Let {T1 , . . . Tu } be an orbit of order u such that p ∤ u and let H = {g ∈ G : gT1 = T1 } be the stabalizer
subgroup of T1 in G.
|G|
We know that u = |OT1 | = [G : GT1 ] = [G : H] = |H|
.

This implies that |G| = pk m = u|H|. We see that pk must divide u|H|. Since p ∤ u then it must be
that pk | |H| and hence pk ≤ |H|

We also know by Lagranges Theorem that |H| | |G|. Since pk | |H| then it must be that p | |H|. Now
suppose that |H| | m then we’d have

m = n|H| = nsp =⇒ p | m

which we know isnt true so it must be that |H| | pk =⇒ |H| ≤ pk . We therefore conclude that |H| = pk
and so Spn has a sylow p-subgroup.

8
 Proposition 5.5 Let H be a subgroup of a finite group G, and suppose that G has a Sylow subgroup
Q. Then H has a Sylow subgroup P. Moreover P = H ∩ xQx−1 for some x ∈ G.
Proof. Let pa be the maximal power of p that divides the order of G. Let pn be the maximal power of p
that divides the order of H. Let us express G as a union of double cosets of H, Q in G

G = ∪Hxi Q

where xi runs through equivalence class representatives. Then we have

|G| |H||Q| |H||Q|

= −1 + · · · + =m
|Q| |H ∩ x1 Qx1 | |H ∩ xk Qx−1
k |

Since p ∤ m then there must exist at least one xi such that |H ∩ xi Qx−1 n
i | has order p . Hence, H has a
Sylow p-subgroup.
Theorem 5.6 A finite group whose order is divisible by p has a Sylow p-subgroup.
Proof. By assumption p | |G| so then the order of G must be of the form |G| = pn m for some positive
interger n ≥ 1. By Cayleys Theorem we know that there exists a permutation group SG that is isomorphic
to G, hence
|SG | = |G| = pn k
Which implies that SG = Spn .

As shown in propostion 5.2 Spn has a subgroup of order pn . A property of isomorphisms is that if one
group has a subgroup of order n then so does the other group. Hence, we conclude since G ∼
= Spn then G
must have a subgroup of order pn . So G has a Sylow p-subgroup.