Direct Product of a Finite Number of Groups

In this session we are going to see about direct product of a finite number of groups.

Definition of Direct product of a finite number of Groups:

Let 𝐺1 , 𝐺2 , . . . , 𝐺𝑛 be a finite collection of groups. The external direct product
of 𝐺1 , 𝐺2 , . . . , 𝐺𝑛 is
𝐺1 ⊕ 𝐺2 ⊕ . . . .⊕ 𝐺𝑛 = {(𝑔1 , 𝑔2 , . . . , 𝑔𝑛 )|𝑔𝑖 ∈ 𝐺𝑖 }
Where (𝑔1 , 𝑔2 , . . . , 𝑔𝑛 )( 𝑔1′ , 𝑔2′ , . . . , 𝑔𝑛′ ) = (𝑔1 𝑔1′ , 𝑔2 𝑔2′ , . . . , 𝑔𝑛 𝑔𝑛′ ) and each
𝑔𝑖 𝑔𝑖′ is performed with the operation of 𝐺𝑖 .

Example:
Suppose 𝐺𝑖 = ℝ (operation addition) for 𝑖 = 1, 2, . . . , 𝑛. Then ℝ × ℝ × . . .× ℝ (n-
factors) is the familiar Euclidean 𝑛 −space ℝ𝑛 with usual vector addition.
(𝑎1 , 𝑎2 , . . . , 𝑎𝑛 ) + (𝑏1 , 𝑏2 , . . . , 𝑏𝑛 ) = (𝑎1 + 𝑏1 , 𝑎2 + 𝑏2 , … , 𝑎𝑛 + 𝑏𝑛 ).

Theorem: Order of an element in a Direct Product

The order of an element in a direct product of a finite number of finite groups is the
least common multiple of the orders of the components of the element. In symbols,
|𝑔1 , 𝑔2 , . . . , 𝑔𝑛 | = 𝑙𝑐𝑚 (|𝑔1 |, |𝑔2 |,. . . , |𝑔𝑛 |).

Proof:
Let the identity of 𝐺𝑖 by 𝑒𝑖 . Let 𝑠 = 𝑙𝑐𝑚 (|𝑔1 |, |𝑔2 |, . . . , |𝑔𝑛 |) and 𝑡 = |𝑔1 , 𝑔2 , … , 𝑔𝑛 |.
Because the fact that 𝑠 is a multiple of each |𝑔𝑖 | implies that
(𝑔1 , 𝑔2 , . . . , 𝑔𝑛 )𝑠 = (𝑔1𝑠 , 𝑔2𝑠 , . . . , 𝑔𝑛𝑠 ) = (𝑒1 , 𝑒2 , . . . , 𝑒𝑛 ), we know that 𝑡 ≤ 𝑠.
On the other hand, from (𝑔1 , 𝑔2 , . . . , 𝑔𝑛 )𝑡 = (𝑔1𝑡 , 𝑔2𝑡 , . . . , 𝑔𝑛𝑡 ) = (𝑒1 , 𝑒2 , . . . , 𝑒𝑛 ),
we see that 𝑡 is a common multiple of |𝑔1 |, |𝑔2 |, . . . , |𝑔𝑛 |. Thus 𝑠 ≤ 𝑡.
Hence 𝑡 = 𝑠 this implies |𝑔1 , 𝑔2 , . . . , 𝑔𝑛 | = 𝑙𝑐𝑚 (|𝑔1 |, |𝑔2 |, . . . , |𝑔𝑛 |).

Example:
Consider the group 𝐺 = ℤ6 × ℤ10 .
(a) What is the order of the element (3,1) in 𝐺?
(b) What is the largest order of an element of 𝐺?

Solution:
a) |(3, 1)| = 𝑙𝑐𝑚 (|3|, |1|) = 𝑙𝑐𝑚(2,10) = 10.
b) Using that|(𝑎, 𝑏)| = 𝑙𝑐𝑚(|𝑎|, |𝑏|) and that the order of any element of ℤ6
divides 6 and the order of any element of ℤ10 divides 10, it follows that the
largest order element of 𝐺 is (1,1) and that|(1,1)| = 𝑙𝑐𝑚(|1|, |1|) =
𝑙𝑐𝑚(6,10) = 30

Example:
How many elements of order 9 does ℤ3 ⊕ ℤ9 have?

Solution:
Let (𝑎, 𝑏) ∈ ℤ3 ⊕ ℤ9 have order 9. Then 𝑙𝑐𝑚 (|𝑎|, |𝑏|) = 9. Since |𝑎| can have
order 1 or 3, then we must have |𝑏| = 9. There are 𝜙(9) = 6 elements of order 9 in
ℤ9 . Then 𝑎 can be any element of ℤ3 , so there are 3 × 6 = 18 elements of the direct
product which have order 9.

Example:
Show that ℤ2 ⊕ ℤ2 ⊕ ℤ2 has seven subgroups of order 2.

Solution:
Each element of order 2 determines a unique subgroup of order 2, hence we only
need to count the elements of order 2. Since|(𝑎, 𝑏, 𝑐)| = 𝑙𝑐𝑚(|𝑎|, |𝑏|, |𝑐|) and the
elements of ℤ2 have order either 1 or 2, then every element of the group will have
order 2 except for the identity (0,0,0). So there are 23 − 1 = 22 elements of order 2.

Theorem:
Let 𝐺 and 𝐻 is finite cyclic groups. Then 𝐺 ⊕ 𝐻 is cyclic if and only if |𝐺| and |𝐻| are
relatively prime.

Proof:
Let |𝐺| = 𝑚 and |𝐻| = 𝑛 so that |𝐺 ⊕ 𝐻| = 𝑚𝑛. To prove the first half of the
theorem we assume 𝐺 ⊕ 𝐻 is cyclic and show that 𝑚 and 𝑛 are relatively prime.
Suppose that 𝑔𝑐𝑑(𝑚, 𝑛) = 𝑑 and (𝑔, ℎ) is a generator of 𝐺 ⊕ 𝐻. Since (𝑔, ℎ)𝑚𝑛/𝑑 =
𝑛 𝑚
((𝑔𝑚 )𝑑 , (ℎ𝑛 ) 𝑑 ) = (𝑒, 𝑒). We have 𝑚𝑛 = |(𝑔, ℎ)| ≤ 𝑚𝑛/𝑑. Thus 𝑑 = 1.
To prove the other half of the theorem let 𝐺 = < 𝑔 > and 𝐻 = < ℎ > and suppose
𝑔𝑐𝑑(𝑚, 𝑛) = 1. Then |(𝑔, ℎ)| = 𝐿𝐶𝑀(𝑚, 𝑛) = 𝑚𝑛 = |𝐺 ⊕ 𝐻| so that (𝑔, ℎ) is a
generator of 𝐺 ⊕ 𝐻.

Internal Direct Product of 𝐻 and 𝐾:

𝐺 is a internal direct product of 𝐻 and 𝐾 if 𝐺 = 𝐻 × 𝐾 where 𝐻 and 𝐾 are normal
subgroups of 𝐺 and 𝐻 ∩ 𝐾 = {𝑒}.

Examples:
 Consider the group (ℤ6 , +) and the following subgroups 𝐻 = {0, 2, 4}, 𝐾 =
{0, 3}. Note that {ℎ ⋆ 𝑘 ∶ ℎ ∈ 𝐻, 𝑘 ∈ 𝐾} = {0, 1, 2, 3, 4, 5} = 𝐺, so the first
condition is met. Also the identity for ℤ6 is 𝑒 = 0 and 𝐻 ∩ 𝐾 = {0} so the
second condition is met. Lastly ℤ6 is an abelian group so the third condition is
met. Which implies (ℤ6 , +) is internal direct product of subgroups 𝐻 =
{0, 2, 4} and 𝐾 = {0, 3}.
 Let 𝑆3 were an internal direct product of its proper subgroups 𝐻 and 𝐾, then
one of the subgroups say 𝐻, would have to have order 3. In this case 𝐻 is the
subgroup of {(1); (123); (132)}. The subgroup 𝐾 must have order 2, but no
matter which subgroup we choose for 𝐾, the condition that ℎ𝑘 = 𝑘ℎ will never
be satisfied for ℎ ∈ 𝐻 and 𝑘 ∈ 𝐾.

Definition of Internal Direct Product of finite number of groups:

Let 𝐻1 , 𝐻2 , . . . , 𝐻𝑛 be a finite collection of normal subgroups of 𝐺. We say that 𝐺 is the
internal direct product of 𝐻1 , 𝐻2 , . . . , 𝐻𝑛 and write 𝐺 = 𝐻1 × 𝐻2 × ⋯ × 𝐻𝑛 , if
1. 𝐺 = 𝐻1 ⋅ 𝐻2 ⋯ 𝐻𝑛 = {ℎ1 ⋅ ℎ2 ⋯ ℎ𝑛 |ℎ𝑖 ∈ 𝐻𝑖 }
2. (𝐻1 ⋅ 𝐻2 ⋯ 𝐻𝑖 ) ∩ 𝐻𝑖+1 = {ℯ} 𝑓𝑜𝑟 𝑖 = 1, 2, . . . , 𝑛 − 1.

Theorem:
If a group 𝐺 is the internal direct product of a finite number of subgroups 𝐻1 , 𝐻2 , … , 𝐻𝑛 ,
then 𝐺 is isomorphic to the external direct product of 𝐻1 , 𝐻2 , … , 𝐻𝑛 . In notation,
𝐻1 × 𝐻2 × ⋯ × 𝐻𝑛 , ≅ 𝐻1 ⨁ 𝐻2 ⨁ … ⨁ 𝐻𝑛 .

Proof:
We first show that the normality of the 𝐻’𝑠 together with the second condition of the
definition guarantees that ℎ’ 𝑠 from different 𝐻𝑖 ’ 𝑠 commute. For if ℎ𝑖 ∈ 𝐻𝑖 and ℎ𝑗 ∈
𝐻𝑗 with 𝑖 ≠ 𝑗, then (ℎ𝑖 ℎ𝑗 ℎ𝑖−1 )ℎ𝑗−1 ∈ 𝐻𝑗 ℎ𝑗−1 = 𝐻𝑗 𝑎𝑛𝑑 ℎ𝑖 (ℎ𝑗 ℎ𝑖−1 ℎ𝑗−1 ) ∈ ℎ𝑖 𝐻𝑖 = 𝐻𝑖
Thus ℎ𝑖 ℎ𝑗 ℎ𝑖−1 ℎ𝑗−1 ∈ 𝐻𝑖 ∩ 𝐻𝑗 = {ℯ} and therefore ℎ𝑖 ℎ𝑗 = ℎ𝑗 ℎ𝑖 . We next claim that each
member of 𝐺 can be expressed uniquely in the form ℎ1 , ℎ2 . . . ℎ𝑛 , where ℎ𝑖 ∈
𝐻𝑖 . That there is at least one such representation is the content of condition 1 of the
definition. To prove uniqueness, suppose that 𝑔 = ℎ1 , ℎ2 . . . ℎ𝑛 and 𝑔 = ℎ1′ ℎ2′ . . . ℎ𝑛′ ,
where ℎ𝑖 and ℎ𝑖′ belong to 𝐻𝑖 for 𝑖 = 1, . . . , 𝑛. Then using the fact that the ℎ’𝑠 from
different 𝐻𝑖 ’s commute we can solve the equation
ℎ1 ℎ2 . . . ℎ𝑛 = ℎ1′ ℎ2′ . . . ℎ𝑛′ for ℎ𝑛′ ℎ𝑛−1 to obtain
ℎ𝑛′ ℎ𝑛−1 = (ℎ1′ )−1 ℎ1 (ℎ2′ )−1 ℎ2 . . . (ℎ𝑛−1
′
)−1 ℎ𝑛−1 .
′ −1
But then ℎ𝑛 ℎ𝑛 ∈ 𝐻1 𝐻2 . . . 𝐻𝑛−1 ∩ 𝐻𝑛 = {𝑒},
So that ℎ𝑛′ ℎ𝑛−1 = 𝑒 and therefore ℎ𝑛′ = ℎ𝑛 . At this point we can cancel ℎ𝑛 and
ℎ𝑛′ from opposite sides of the equal sign in equation (1) and repeat the preceding
′
argument to obtain ℎ𝑛−1 = ℎ𝑛−1 . Continuing in this fashion, we eventually have
′
ℎ𝑖 = ℎ𝑖 for 𝑖 = 1, . . . , 𝑛. With our claim established, we may now define a function 𝜙
from 𝐺 to 𝐻1 ⨁ 𝐻2 ⨁ … ⨁ 𝐻𝑛 by 𝜙(ℎ1 ℎ2 . . . ℎ𝑛 ) = (ℎ1 ℎ2 . . . ℎ𝑛 ). We leave to the
reader the easy verification that 𝜙 is an isomorphism.

Theorem:
Every group of order 𝑝2 , where 𝑝 is a prime is isomorphic to ℤ𝑝2 or ℤ𝑝 ⨁ ℤ𝑝 .

Proof:
Let 𝐺 be a group of order 𝑝2 , where 𝑝 is a prime. If 𝐺 has an element of order 𝑝2 ,
then 𝐺 is isomorphic to ℤ𝑝2 . So we may assume that every nonidentity element
of 𝐺 has order 𝑝. First we show that for any element 𝑎, the subgroup < 𝑎 > is normal
in 𝐺. If this is not the case then there is an element 𝑏 in 𝐺 such that 𝑏𝑎𝑏 −1 is not
in < 𝑎 >. Then < 𝑎 > and < 𝑏𝑎𝑏 −1 > are distinct subgroups of order 𝑝. Since < 𝑎 > ∩
< 𝑏𝑎𝑏 −1 > is a subgroup of both < 𝑎 > and < 𝑏𝑎𝑏 −1 >, we have that < 𝑎 > ∩ <
𝑏𝑎𝑏 −1 > = {𝑒} . From this it follows that the distinct left cosets of < 𝑏𝑎𝑏 −1 > are <
𝑏𝑎𝑏 −1 > , 𝑎 < 𝑏𝑎𝑏 −1 > , 𝑎2 < 𝑏𝑎𝑏 −1 > , . . . , 𝑎𝑝−1 < 𝑏𝑎𝑏 −1 >. Since 𝑏 −1 must lie in one
of these cosets, we may write 𝑏 −1 in the form 𝑏 −1 for some 𝑖 and 𝑗. Canceling
the 𝑏 −1 terms, we obtain 𝑒 = 𝑎𝑖 𝑏𝑎 𝑗 and therefore 𝑏 = 𝑎−𝑖−𝑗 ∈ < 𝑎 >. This
contradiction verifies our assertion that every subgroup of the form < 𝑎 > is normal in
𝐺. To complete the proof let 𝑥 be any non identity element in 𝐺 and 𝑦 be any element
of 𝐺 not in < 𝑥 >. Then by comparing orders we see that 𝐺 = < 𝑥 > × < 𝑦 > ≅ <
ℤ𝑝 ⨁ ℤ𝑝 .

Theorem:
Let 𝐺1 , 𝐺2 , . . . , 𝐺𝑛 be groups and let 𝐺 = 𝐺1 × 𝐺2 × ⋯ × 𝐺𝑛 be their direct product.
Then
 a) For each fixed 𝑖 the set of elements of 𝐺 which have the identity of 𝐺𝑗 in the 𝑗 𝑡ℎ
position for all 𝑗 ≠ 𝑖 and arbitrary elements of 𝐺𝑖 in position 𝑖 is a subgroup
of 𝐺 isomorphic to 𝐺𝑖 . In notation,
𝐺𝑖 ≅ { (1, 1, . . . , 1, 𝑔𝑖 , 1, . . . , 1)| 𝑔𝑖 ∈ 𝐺𝑖 },
𝐺
If 𝐺𝑖 with this subgroup, then 𝐺𝑖 ⊴ 𝐺 and ≅ 𝐺1 × 𝐺2 × ⋯ × 𝐺𝑖−1 × 𝐺𝑖+1 ×
𝑖𝐺
⋯ × 𝐺𝑛
b) Under the identifications in part (𝑎) if 𝑥 ∈ 𝐺𝑖 and 𝑦 ∈ 𝐺𝑗 for some 𝑖 ≠ 𝑗,
then 𝑥𝑦 = 𝑦𝑥.

Proof:
a) Since the operation in G is defined componentwise, it follows easily from the
subgroup criterion that {(1, 1, . . . , 1, 𝑔𝑖 , 1, . . . , 1)| 𝑔𝑖 ∈ 𝐺𝑖 } is a subgroup of 𝐺.
Furthermore, the map 𝑔𝑖 ↦ (1, 1, . . . , 1, 𝑔𝑖 , 1, . . . , 1) is seen to be an
isomorphism of 𝐺𝑖 , with this subgroup. Identify 𝐺𝑖 , with this isomorphic copy
in 𝐺.
To prove the remaining parts of (𝑎) consider the map 𝜑: 𝐺 → 𝐺1 × 𝐺2 × ⋯ ×
𝐺𝑖−1 × 𝐺𝑖+1 × ⋯ × 𝐺𝑛 defined by,
𝜑(𝑔1 , 𝑔2 , … , 𝑔𝑛 ) = (𝑔1 , 𝑔2 , … 𝑔𝑖−1 , 𝑔𝑖+1 , … , 𝑔𝑛 )
The map 𝜑 is a homomorphism since
𝜑((𝑔1 , 𝑔2 , … , 𝑔𝑛 )( ℎ1 , ℎ2 , … , ℎ𝑛 )) = 𝜑(𝑔1 ℎ1 , 𝑔2 ℎ2 , … , 𝑔𝑛 ℎ𝑛 )
= (𝑔1 ℎ1 , 𝑔2 ℎ2 , … , 𝑔𝑖−1 ℎ𝑖−1 , 𝑔𝑖+1 ℎ𝑖+1 , … , 𝑔𝑛 ℎ𝑛 )
= (𝑔1 , 𝑔2 , … , 𝑔𝑖−1 , 𝑔𝑖+1 , … , 𝑔𝑛 ) (ℎ1 , ℎ2 , … , ℎ𝑖−1 , ℎ𝑖+1 , … , ℎ𝑛 )
= 𝜑(𝑔1 , 𝑔2 , … , 𝑔𝑛 ) 𝜑(ℎ1 , ℎ2 , … , ℎ𝑛 ).
Since the entries in position 𝑗 are arbitrary elements of 𝐺𝑗 for all 𝑗, 𝜑 is
surjective. Furthermore,
𝐾𝑒𝑟 𝜑 = {(𝑔1 , 𝑔2 , . . . , 𝑔𝑛 ) | 𝑔𝑖 = 1 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑗 ≠ 𝑖} = 𝐺𝑖 .
This proves that 𝐺𝑖 is a normal subgroup of 𝐺 (in particular, it again proves
this copy of 𝐺, is a subgroup) and the First Isomorphism Theorem gives the
final assertion of part (a).
b) If 𝑥 = (1, … , 1, 𝑔𝑖 , 1, … , 1 ) and 𝑦 = (1, … , 1, 𝑔𝑗 , 1, … , 1 ) where the indicated
entries appear in positions 𝑖, 𝑗 respectively then 𝑥𝑦 =
(1, … ,1, 𝑔𝑖 , 1, … , 1, 𝑔𝑗 , 1, … , 1) = 𝑦𝑥, where 𝑖 < 𝑗.
Hence Proved.